diff --git a/AGENTS.md b/AGENTS.md
index f10fb39483c..d468f7d1984 100644
--- a/AGENTS.md
+++ b/AGENTS.md
@@ -46,7 +46,7 @@ Repository map and Reference Documentation sections below.
## Project at a Glance
-**OmniRoute** — unified AI proxy/router. One endpoint, 348 LLM providers, auto-fallback.
+**OmniRoute** — unified AI proxy/router. One endpoint, 349 LLM providers, auto-fallback.
| Layer | Location | Purpose |
| ------------- | ----------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
diff --git a/PROVIDER_REFERENCE.md b/PROVIDER_REFERENCE.md
new file mode 100644
index 00000000000..571fe0e904c
--- /dev/null
+++ b/PROVIDER_REFERENCE.md
@@ -0,0 +1,447 @@
+---
+title: "Provider Reference"
+version: 3.8.50
+lastUpdated: 2026-08-21
+---
+
+# Provider Reference
+
+> **Auto-generated** from `src/shared/constants/providers.ts` — do not edit by hand.
+> Regenerate with: `npm run gen:provider-reference`
+> **Last generated:** 2026-08-21
+
+Total providers: **349**. See category breakdown below.
+
+## Categories
+
+- **Free** — free tier with API key (configured via dashboard)
+- **No-auth** — public endpoints that require no key or sign-in at all
+- **OAuth** — sign-in flow handled by OmniRoute, no API key needed
+- **Web cookie** — wraps the provider's web app via cookie auth
+- **API key** — paid provider configured via API key (free credits may apply)
+- **Local** — runs on the user's machine (Ollama, LM Studio, vLLM, etc.)
+- **Search** — web search providers
+- **Audio** — audio-only providers (TTS/STT)
+- **Upstream proxy** — providers that proxy to other providers
+- **Cloud agent** — long-running coding agents (Codex Cloud, Devin, Jules)
+- **System** — OmniRoute-internal providers (loopback, etc.)
+
+Additional tags: `image`, `video`, `aggregator`, `enterprise`, `embed/rerank`, `self-hosted`.
+
+`Tool calling` (where shown): `native` — real function-calling API; `emulated` — the `tools` array is prompt-emulated via `webTools.ts` (regex-parsed `{...}` blocks); `none` — `tools` is currently silently dropped. See #7286.
+
+Use the dashboard at `/dashboard/providers` to enable, configure, and test each provider.
+
+---
+
+## No-auth Providers (no key required) (11)
+
+| ID | Alias | Name | Tags | Website | Notes | Tool calling |
+|----|-------|------|------|---------|-------|--------------|
+| `aihorde` | `horde` | AI Horde | No-auth | [link](https://aihorde.net) | No API key required — uses AI Horde's documented anonymous key. Adding a free aihorde.net key is optional and only buys higher queue priority (kudos). | — |
+| `auggie` | `aug` | Augment (Auggie CLI) | No-auth | [link](https://augmentcode.com) | No API key stored by OmniRoute. Install the Auggie CLI and run `auggie login` on this machine, then OmniRoute spawns it locally for each request. | — |
+| `chipotle` | `pepper` | Chipotle Pepper AI (Free) | No-auth | [link](https://amelia.chipotle.com) | No credentials required. Uses Chipotle's public support chatbot via reverse-engineered SockJS/STOMP protocol. | — |
+| `cloudflare-playground` | `cfp` | Cloudflare AI Playground | No-auth | [link](https://playground.ai.cloudflare.com) | No credentials required — anonymous browser sessions over a reverse-engineered cf_agent WebSocket protocol (Playwright transport). | — |
+| `devin-cli-agentic` | `dva` | Devin CLI Agentic Bridge | No-auth | [link](https://docs.devin.ai/work-with-devin/devin-cli) | Authentication is owned by the official Devin CLI in its isolated bridge volume. | emulated |
+| `duckduckgo-web` | `ddgw` | DuckDuckGo AI Chat | No-auth | [link](https://duckduckgo.com/duckchat) | No credentials required — DuckDuckGo AI Chat is anonymous and free. | emulated |
+| `felo-web` | `felo` | Felo | No-auth | [link](https://felo.ai) | No credentials required — Felo is a free, no-signup chat/search aggregator. | — |
+| `opencode` | `oc` | OpenCode Free | No-auth | [link](https://opencode.ai) | No API key required — uses OpenCode's public free endpoint. | — |
+| `theoldllm` | `tllm` | The Old LLM (Free) | No-auth | [link](https://theoldllm.vercel.app) | No credentials required. The executor auto-generates access tokens via an embedded Playwright browser instance. | — |
+| `veoaifree-web` | `veo-free` | Veo AI Free | No-auth, video | [link](https://veoaifree.com) | No auth required. Rate limited to 6 requests/hour per IP. | — |
+| `zcode` | `zc` | ZCode (GLM Coding Plan) | No-auth | [link](https://zcode.z.ai) | No API key stored by OmniRoute. The local ZCode app-server uses the existing builtin:zai-coding-plan login. | — |
+
+## OAuth Providers (25)
+
+| ID | Alias | Name | Tags | Website | Notes |
+|----|-------|------|------|---------|-------|
+| `agy` | `agy` | Antigravity CLI | OAuth | [link](https://antigravity.google) | Import your Antigravity CLI (`agy`) login (paste/upload its token file), auto-detect a local CLI login, or sign in with Google. Shares the Antigravity backend (incl. Claude models). |
+| `amazon-q` | `aq` | Amazon Q | OAuth | [link](https://aws.amazon.com/q/developer/) | Uses the same AWS Builder ID or imported refresh-token flow as Kiro, but keeps Amazon Q connections separate. |
+| `antigravity` | — | Antigravity | OAuth | — | — |
+| `claude` | `cc` | Claude Code | OAuth | — | — |
+| `cline` | `cl` | Cline | OAuth | — | — |
+| `clinepass` | `cp` | ClinePass | OAuth | [link](https://cline.bot/cline-pass) | ClinePass is Cline's $9.99/mo subscription bundling 10 open coding models. Sign in with your Cline account (same login as the Cline CLI/IDE), or paste a direct ClinePass API key (app.cline.bot → Settings → API Keys). A ClinePass subscription unlocks the cline-pass/* models. Reuses the Cline WorkOS OAuth flow. |
+| `codebuddy-cn` | `cbcn` | CodeBuddy CN | OAuth | [link](https://copilot.tencent.com) | Tencent CodeBuddy CN (copilot.tencent.com). Sign in via the official CLI device-code flow, or paste a direct API key (sent as Authorization: Bearer). Catalog: GLM / Kimi / MiniMax / DeepSeek / Hunyuan. |
+| `codex` | `cx` | OpenAI Codex | OAuth | — | — |
+| `cursor` | `cu` | Cursor IDE | OAuth | — | — |
+| `devin-cli` | `dv` | Devin CLI | OAuth | [link](https://cli.devin.ai) | Requires the Devin CLI binary. Run `devin auth login` to authenticate, or provide your WINDSURF_API_KEY. Install: https://cli.devin.ai |
+| `devin-desktop` | — | Devin Desktop | OAuth | [link](https://devin.ai) | Paste an existing Devin API key from an authenticated Devin session. Key export availability and steps vary by Devin version and account. |
+| `ghe-copilot` | `ghe-copilot` | GitHub Enterprise Copilot | OAuth | — | Enter your GHE instance URL (e.g., https://ghe.company.com) in provider settings, then authenticate via device flow. |
+| `github` | `gh` | GitHub Copilot | OAuth | — | — |
+| `gitlab-duo` | `gitlab-duo` | GitLab Duo | OAuth | [link](https://docs.gitlab.com/user/duo_agent_platform/code_suggestions/) | GitLab Duo OAuth is not configured. Register an OAuth application at https://gitlab.com/-/profile/applications with redirect URI http://localhost:20128/callback and scopes "ai_features read_user", then set GITLAB_DUO_OAUTH_CLIENT_ID (and optionally GITLAB_DUO_OAUTH_CLIENT_SECRET) and restart. |
+| `grok-cli` | `gc` | Grok Build | OAuth | — | Sign in with your browser, or paste your ~/.grok/auth.json (or the JWT access token) from the Grok Build CLI; refresh_token is rotated automatically either way. |
+| `kilocode` | `kc` | Kilo Code | OAuth | — | — |
+| `kimi-coding` | `kmc` | Kimi Code CLI | OAuth | [link](https://www.kimi.com/code?aff=omniroute) | Sign in with the same Kimi account used by Kimi Code CLI. OmniRoute uses the CLI OAuth flow and Kimi Coding Plan endpoints. |
+| `kiro` | `kr` | Kiro AI | OAuth | — | Free tier: 50 credits/month (~25K–100K tokens). ⚠️ Kiro ToS prohibits third-party proxy/harness use. |
+| `openference` | `of` | Openference | OAuth | [link](https://openference.com) | Sign in with your Openference account to route requests through api.openference.com. An active plan is required for inference — OAuth may authenticate but return 402 without one. |
+| `qoder` | `if` | Qoder | OAuth | — | — |
+| `raycast` | `rc` | Raycast Pro AI | OAuth | [link](https://raycast.com/ai) | Unofficial integration — uses your Raycast Pro subscription via credentials from the macOS app (Auto-Import or manual capture). May break on Raycast updates. Not for redistribution; personal use only. |
+| `trae` | `tr` | Trae | OAuth | [link](https://trae.ai) | Trae is an AI-native IDE by ByteDance (SOLO remote agent). Authorize via trae.ai in the popup, or sign in at solo.trae.ai and paste the Cloud-IDE-JWT (sent as 'Authorization: Cloud-IDE-JWT ', ~14-day lifetime) as the access token; web_id/biz_user_id/user_unique_id/scope/tenant/region propagate via providerSpecificData. No headless refresh for pasted tokens — re-paste on expiry. |
+| `xai-oauth` | `xao` | xAI OAuth (Grok) | OAuth | [link](https://x.ai) | Sign in with xAI to use api.x.ai models such as Grok 4.5. This is separate from Grok Build JWT sessions, which use cli-chat-proxy.grok.com and grok-build model aliases. |
+| `zed` | `zd` | Zed IDE | OAuth | [link](https://zed.dev) | Zed stores LLM provider credentials (OpenAI, Anthropic, Google, Mistral, xAI) in the OS keychain. Use the Import button below to discover and import them automatically. |
+| `zed-hosted` | — | Zed Hosted Models | OAuth | [link](https://zed.dev) | Sign in with your Zed account (native-app sign-in). OmniRoute generates a one-time RSA keypair and opens zed.dev to authorize it — on a remote/headless install, copy the resulting 127.0.0.1 callback URL from your browser's address bar and paste it back here. Distinct from the 'Zed IDE' credential-import entry above: this proxies chat completions through Zed's own hosted model aggregator (cloud.zed.dev), fronting Anthropic/OpenAI/Google/xAI models under your Zed plan. |
+
+## Web Cookie Providers (35)
+
+| ID | Alias | Name | Tags | Website | Notes | Tool calling |
+|----|-------|------|------|---------|-------|--------------|
+| `adapta-web` | `adp-web` | Adapta.org (Adapta One Web) | Web cookie | [link](https://agent.adapta.one) | Paste your __client cookie value from .clerk.agent.adapta.one (DevTools → Application → Cookies) | emulated |
+| `adobe-firefly` | `firefly` | Adobe Firefly (Image/Video) | Web cookie | [link](https://firefly.adobe.com) | RECOMMENDED: firefly.adobe.com signed-in → F12 → Network → click firefly-3p.ff.adobe.io (generate-async or models/discovery) → Request Headers → Authorization → copy the token AFTER 'Bearer ' (starts with eyJ…). Cookie-only from firefly.adobe.com mints a GUEST token → 401/403; only multi-domain IMS cookies (adobelogin.com) or that Bearer JWT work. Unofficial/experimental media + Limits. | — |
+| `blackbox-web` | `bb-web` | Blackbox Web (Subscription) | Web cookie | [link](https://app.blackbox.ai) | Paste your __Secure-authjs.session-token value or full cookie header from app.blackbox.ai | emulated |
+| `chatgpt-web` | `cgpt-web` | ChatGPT Web (Plus/Pro) | Web cookie | [link](https://chatgpt.com) | Paste your __Secure-next-auth.session-token cookie value from chatgpt.com | emulated |
+| `chatgpt-web-codex` | `cgpt-codex` | ChatGPT Web (Codex) | Web cookie | [link](https://chatgpt.com) | Paste the full ChatGPT Cookie header. OmniRoute verifies it in an isolated headless browser profile. | native |
+| `claude-web` | `cw` | Claude Web | Web cookie | [link](https://claude.ai) | Paste your session cookie from claude.ai | none |
+| `conol-web` | `cnl` | Conol (Unofficial/Experimental) | Web cookie | [link](https://conol.ai) | Use browser sign-in, or paste the full Cookie header from conol.ai. The __Secure-better-auth.session_token cookie is required. | — |
+| `copilot-m365-web` | `m365copilot` | Microsoft 365 Copilot (BizChat) | Web cookie | [link](https://m365.cloud.microsoft/chat) | Sign in at m365.cloud.microsoft/chat, then open DevTools → Network → filter 'WS' → click the Chathub WebSocket connection. Copy both the access_token query parameter AND the account-specific Chathub path segment from its request URL (wss://…/Chathub/?…&access_token=…). It is NOT an Authorization: Bearer header on an XHR/Fetch request. The token is short-lived; this is an unofficial integration. Optional: store a refresh_token in providerSpecificData.refreshToken (any Microsoft device-code/refresh flow for the substrate.office.com/sydney scopes) and OmniRoute pre-flight-refreshes the access token itself — otherwise re-capture after every ~75 min expiry. | — |
+| `copilot-web` | `copilot` | Microsoft Copilot Web | Web cookie | [link](https://copilot.microsoft.com) | Paste the access_token from an authenticated copilot.microsoft.com request (DevTools → Network → Authorization), or export a HAR while logged in | — |
+| `deepseek-web` | `ds-web` | DeepSeek Web | Web cookie | [link](https://chat.deepseek.com) | Paste your userToken from chat.deepseek.com — DevTools → Application → Local Storage → userToken | emulated |
+| `doubao-web` | `db` | Dola Web (ByteDance) | Web cookie | [link](https://www.dola.com) | Paste the full Cookie header from www.dola.com. It should include sessionid, ttwid, and s_v_web_id. If s_v_web_id is unavailable, fp=verify_... from a chat/completion request URL can be used as a fallback. | — |
+| `gemini-business` | `gembiz` | Gemini Business (Enterprise) | Web cookie | [link](https://business.gemini.google) | From your enterprise account: open business.gemini.google/home/cid/{your-cid}, then copy __Secure-1PSID and __Secure-1PSIDTS cookies from DevTools → Application → Cookies. Paste as a cookie header below. | — |
+| `gemini-web` | `gweb` | Gemini Web (Free) | Web cookie | [link](https://gemini.google.com) | Paste your __Secure-1PSID cookie value from gemini.google.com. Optionally add __Secure-1PSIDTS separated by semicolon. | emulated |
+| `grok-web` | `gw` | Grok Web (Subscription) | Web cookie | [link](https://grok.com) | Paste the full grok.com cookie line from DevTools → Application → Cookies. Include both `sso` and `sso-rw` (e.g. `sso=...; sso-rw=...`) — Grok's anti-bot rejects `sso` on its own. | — |
+| `hailuo-web` | `hailuo-web` | Hailuo Web (MiniMax) | Web cookie | [link](https://hailuo.ai) | Open hailuo.ai, log in, then open DevTools → Application → Local Storage → copy the "_token" value. device_id/uuid fingerprint fields are derived automatically; if requests fail, re-capture _token (sessions can expire). | — |
+| `huggingchat` | `huggingchat` | HuggingChat (Free) | Web cookie | [link](https://huggingface.co/chat) | Paste the full Cookie header from huggingface.co/chat (DevTools → Network → /chat/conversation → Request Headers → Cookie). It should include hf-chat and may also include token / aws-waf-token. | — |
+| `hyperagent` | `ha` | HyperAgent (Unofficial/Experimental) | Web cookie | [link](https://hyperagent.com) | Paste the full Cookie header from hyperagent.com (DevTools → Network → any request → Request Headers → Cookie). Session cookies power chat + billing usage. | — |
+| `inner-ai` | `in-ai` | Inner.ai (Subscription) | Web cookie | [link](https://app.innerai.com) | Paste your token cookie and email separated by a space: open DevTools → Application → Cookies → .innerai.com, copy the token value, then append a space and your Inner.ai login email. Example: eyJhbG... user@example.com | emulated |
+| `kimi-web` | `kimi-web` | Kimi Web | Web cookie | [link](https://www.kimi.com/code?aff=omniroute) | Paste access_token from www.kimi.com DevTools → Application → Local Storage. A legacy kimi-auth cookie is also accepted. | — |
+| `lmarena` | `lma` | Arena (Free) | Web cookie | [link](https://arena.ai) | Paste the full Cookie header from arena.ai (DevTools → Network → request → Cookie). Include arena-auth-prod-v1.0/.1… and cf_clearance/__cf_bm when present. OmniRoute uses Chrome TLS impersonation; if Arena still 403s, set providerSpecificData.recaptchaV3Token from a live browser session. | — |
+| `microsoft-designer-web` | `msdesigner` | Microsoft Designer (Image Generation) | Web cookie | [link](https://designer.microsoft.com) | Sign in at designer.microsoft.com, then open DevTools → Network, generate an image, and find the request to DallE.ashx?action=GetDallEImagesCogSci. Copy the value of its Authorization: Bearer header (the access_token — no 'Bearer ' prefix). The token is short-lived; this is an unofficial, reverse-engineered integration. | — |
+| `muse-spark-web` | `ms-web` | Muse Spark Web (Meta AI) | Web cookie | [link](https://www.meta.ai) | Paste your ecto_1_sess cookie AND the ecto1:... WS auth token from meta.ai. Capture the ecto1: token in DevTools → Network → WS → the clippy request's Authorization query param. Example: ecto_1_sess=4240a308...NVDg0; ecto1:ABCD... | emulated |
+| `notion-web` | `nw` | Notion AI Web (Unofficial/Experimental) | Web cookie | [link](https://www.notion.so) | Paste only the token_v2 cookie VALUE from app.notion.com (DevTools → Application → Cookies → token_v2). Do not paste token_v2= or the full Cookie header. Workspace is auto-detected; space_id / notion_user_id are optional. | — |
+| `perplexity-web` | `pplx-web` | Perplexity Web (Pro/Max) | Web cookie | [link](https://www.perplexity.ai) | Paste your __Secure-next-auth.session-token cookie value from perplexity.ai | emulated |
+| `poe-web` | `poe` | Poe Web (Subscription) | Web cookie | [link](https://poe.com) | Paste your p-b cookie value from poe.com (DevTools → Application → Cookies → p-b) | — |
+| `promptql` | `pql` | PromptQL (Unofficial/Experimental) | Web cookie | [link](https://prompt.ql.app) | Paste the Bearer JWT from prompt.ql.app DevTools → Network → graphql → Authorization (token only). Optional projectId + session Cookie for refresh. | — |
+| `qwen-web` | `qwen-web` | Qwen Web (Free) | Web cookie | [link](https://chat.qwen.ai) | Open chat.qwen.ai, log in, then open DevTools → Application → Local Storage → copy the "token" value (or use tongyi_sso_ticket cookie as Bearer token). | emulated |
+| `t3-web` | `t3chat` | t3.chat (Pro/Free) | Web cookie | [link](https://t3.chat) | Open t3.chat in your browser, log in, then open DevTools → Application → Local Storage → https://t3.chat. Copy the value of 'convex-session-id'. Also open DevTools → Network, copy the Cookie header from any request. Paste both values here. See provider setup docs for a step-by-step guide. | emulated |
+| `tencent-aistudio-web` | `tasw` | Tencent AI Studio (Free) | Web cookie | [link](https://aistudio.tencent.ai) | Log in to aistudio.tencent.ai, open DevTools -> Network, copy any request Cookie header containing session tokens. | — |
+| `tinycms-web` | `tcw` | TinyCMS Web (Free/Sub) | Web cookie | [link](https://site.tinycms.xyz) | Go to site.tinycms.xyz, open DevTools → Application → Local Storage, copy the value of 'app-config-uuid' (starts with 'R'), and paste it here. | — |
+| `v0-vercel-web` | `v0-vercel-web` | v0 Vercel Web (Code Gen) | Web cookie | [link](https://v0.dev) | Paste your session cookie from v0.dev (DevTools → Application → Cookies) | — |
+| `venice-web` | `ven` | Venice Web (Privacy) | Web cookie | [link](https://venice.ai) | Paste your session cookie from venice.ai (DevTools → Application → Cookies) | — |
+| `yuanbao-web` | `ybw` | Tencent Yuanbao (Free) | Web cookie | [link](https://yuanbao.tencent.com) | Log in to yuanbao.tencent.com, then paste the full Cookie header (DevTools → Network → any /api request → Request Headers → Cookie). It must contain hy_user and hy_token. | — |
+| `zai-web` | `zw` | Z.ai Web | Web cookie | [link](https://chat.z.ai) | Copy the "token" value from chat.z.ai → DevTools → Application → Local Storage. Do not copy cookies; OmniRoute handles the per-request CAPTCHA through its browser transport. | — |
+| `zenmux-free` | `zmf` | ZenMux Free (Web) | Web cookie | [link](https://zenmux.ai) | Login at zenmux.ai, then export all cookies using EditThisCookie or Cookie-Editor and paste the full Cookie header string here. Refresh every ~30 days. | — |
+
+## API Key Providers (paid / paid-with-free-credits) (233)
+
+| ID | Alias | Name | Tags | Website | Notes |
+|----|-------|------|------|---------|-------|
+| `360ai` | `360ai` | 360 AI | API key | [link](https://ai.360.cn) | Get API key at ai.360.cn |
+| `agentrouter` | `agentrouter` | AgentRouter | API key, aggregator | [link](https://agentrouter.org) | $200 free credits on signup - multi-model routing gateway |
+| `agnes` | `agnes` | Agnes AI | API key, video | [link](https://agnes-ai.com) | Get API key at agnes-ai.com |
+| `ai21` | `ai21` | AI21 Labs | API key | [link](https://www.ai21.com) | $10 trial credits on signup (valid 3 months), no credit card required |
+| `aimlapi` | `aiml` | AI/ML API | API key, aggregator | [link](https://aimlapi.com) | Free tier paused (2026) — AI/ML API is now pay-as-you-go only (min $20 top-up); no recurring free credits. |
+| `ainative` | `ainative` | AINative Studio | API key | [link](https://ainative.studio) | Create a free API key at ainative.studio (no card), then paste it here as a Bearer token. |
+| `aion` | `aion` | Aion Labs | API key | [link](https://www.aionlabs.ai) | Create a free API key at aionlabs.ai (no card), then paste it here as a Bearer token. |
+| `alibaba` | `ali` | Alibaba Cloud Model Studio | API key | [link](https://bailian.console.alibabacloud.com/) | — |
+| `alibaba-cn` | `ali-cn` | Alibaba (China) | API key | [link](https://dashscope.console.aliyun.com/) | — |
+| `ant-ling` | `ling` | Ant Ling / Ring (inclusionAI) | API key | [link](https://developer.ant-ling.com/en/docs/) | Register and create an API key at the Ant Ling API console (https://chat.ant-ling.com/open), then paste it here. OmniRoute routes chat traffic to https://api.ant-ling.com/v1/chat/completions; the provider is OpenAI-compatible and also exposes an Anthropic-compatible surface. |
+| `anthropic` | `anthropic` | Anthropic | API key | [link](https://platform.claude.com) | — |
+| `anyapi` | `anyapi` | AnyAPI AI | API key, aggregator | [link](https://anyapi.ai) | Free plan: 100,000 ANY Tokens/day and 100 RPM for eligible Free/Basic models; no credit card required. |
+| `api-airforce` | `af` | Api.airforce | API key | [link](https://api.airforce) | 55 free tier models including Grok-3, Claude 3.7, Qwen3, Kimi-K2, Gemini 2.5 Flash, DeepSeek-V3 |
+| `arcee-ai` | `arcee` | Arcee AI | API key | [link](https://arcee.ai) | Get API key at arcee.ai |
+| `auriko` | `auriko` | Auriko | API key, aggregator | [link](https://www.auriko.ai) | Free plan publishes 1,000 Platform RPM and 10,000 BYOK RPM. Platform inference still passes through provider cost; this is not a free-token pool or unlimited free inference. |
+| `azure-ai` | `azure-ai` | Azure AI Foundry | API key, enterprise | [link](https://learn.microsoft.com/azure/ai-foundry) | Use your Azure AI Foundry key. Base URL can be https://.services.ai.azure.com/openai/v1/ or https://.openai.azure.com/openai/v1/. |
+| `azure-openai` | `azure` | Azure OpenAI | API key, enterprise | [link](https://azure.microsoft.com/products/ai-services/openai-service) | Use your Azure OpenAI API key. Base URL should be your resource endpoint, for example https://my-resource.openai.azure.com. |
+| `bai` | `bai` | b.ai | API key | [link](https://b.ai) | Bearer API key for the b.ai OpenAI-compatible LLM gateway (distinct from TheB.AI). Create a key at https://docs.b.ai, then use https://api.b.ai/v1 as the OpenAI-compatible base URL. |
+| `baichuan` | `baichuan` | Baichuan | API key | [link](https://www.baichuan-ai.com/) | Get API key at platform.baichuan-ai.com |
+| `baidu` | `baidu` | Baidu (ERNIE) | API key | [link](https://ernie.baidu.com/) | Get API key at console.bce.baidu.com |
+| `bailian-coding-plan` | `bcp` | Alibaba Token Plan | API key | [link](https://www.alibabacloud.com/help/en/model-studio/token-plan-overview) | — |
+| `baseten` | `baseten` | Baseten | API key | [link](https://baseten.co) | $30 free trial credits for GPU inference |
+| `bazaarlink` | `bzl` | BazaarLink | API key | [link](https://bazaarlink.ai) | Use your BazaarLink API key (starts with sk-bl-) in Authorization: Bearer . OpenAI SDK works with base URL https://bazaarlink.ai/api/v1. Models use provider/model-name format. |
+| `bedrock` | `bedrock` | Amazon Bedrock | API key, enterprise | [link](https://aws.amazon.com/bedrock) | Use your Amazon Bedrock API key and configure the AWS region where your models are enabled (for example eu-west-2). OmniRoute calls Bedrock's native Converse API directly. |
+| `black-forest-labs` | `bfl` | Black Forest Labs | API key, image | [link](https://blackforestlabs.ai) | — |
+| `blackbox` | `bb` | Blackbox AI | API key | [link](https://blackbox.ai) | Limited free access is available through Blackbox; model availability and account limits apply |
+| `bluesminds` | `bm` | BluesMinds | API key | [link](https://www.bluesminds.com) | Free daily pi credits — supports 200+ models including GPT-4o, GPT-4.1, Claude Sonnet 4.5, Gemini 2.0 Flash, DeepSeek V4, Qwen, Kimi K2 |
+| `byteplus` | `bpm` | BytePlus ModelArk | API key | [link](https://console.byteplus.com/ark) | — |
+| `bytez` | `bytez` | Bytez | API key | [link](https://bytez.com) | $1 free credits, refreshes every 4 weeks |
+| `cerebras` | `cerebras` | Cerebras | API key | [link](https://inference.cerebras.ai) | Free Trial: 1M tokens/day, 30K TPM, 5 RPM — no credit card. |
+| `charm-hyper` | `charm-hyper` | Charm Hyper | API key | [link](https://hyper.charm.land) | 100 free monthly Hypercredits on signup |
+| `chat-oripe` | `chat-oripe` | Chat Oripe | API key, aggregator | [link](https://api.oriper.com) | Official metadata advertises 2M tokens/month, but the public site and documentation were blocked during audit; treat the quota and brand mapping as unconfirmed. |
+| `chatanywhere` | `chatanywhere` | ChatAnywhere | API key, aggregator | [link](https://chatanywhere.tech) | Personal, educational or research use only: public documentation cites 10,000 points/day and 200 requests/day per IP/key; do not use for commercial traffic. |
+| `cheaperinference` | `cinf` | Cheaper Inference | API key | [link](https://cheaperinference.com/?utm_source=omniroute) | — |
+| `chenzk` | `chenzk` | Chenzk API | API key | [link](https://chenzk.top) | — |
+| `chutes` | `chutes` | Chutes.ai | API key, aggregator | [link](https://chutes.ai) | Bearer API key for the Chutes OpenAI-compatible gateway. |
+| `clarifai` | `clarifai` | Clarifai | API key, enterprise | [link](https://docs.clarifai.com) | Use your Clarifai PAT or app-specific API key. OmniRoute targets the OpenAI-compatible endpoint at https://api.clarifai.com/v2/ext/openai/v1 and authenticates with Authorization: Key . |
+| `cloudcode-one` | `cloudcode-one` | CloudCode.ONE | API key, aggregator | [link](https://cloudcode.one) | Published free models include glm-4.7-flash and glm-4.6v-flash; no numeric quota is published, and key creation may require credit or a coupon. |
+| `cloudflare-ai` | `cf` | Cloudflare Workers AI | API key | [link](https://developers.cloudflare.com/workers-ai) | Requires API Token AND Account ID (found at dash.cloudflare.com) |
+| `clova-studio` | `clova` | Naver CLOVA Studio | API key | [link](https://api.ncloud-docs.com/docs/en/ai-naver-clovastudio-summary) | — |
+| `codestral` | `codestral` | Codestral | API key | [link](https://mistral.ai) | — |
+| `cohere` | `cohere` | Cohere | API key | [link](https://cohere.com) | Free Trial: 1,000 API calls/month for testing, no credit card required |
+| `command-code` | `cmd` | Command Code | API key | [link](https://commandcode.ai/) | Use a Command Code API key. Requests are sent to Command Code's /alpha/generate endpoint. |
+| `coze` | `coze` | Coze | API key | [link](https://coze.com) | Get API key at coze.com/open/api |
+| `crof` | `crof` | CrofAI | API key | [link](https://crof.ai) | — |
+| `cursor-api` | `cua` | Cursor API | API key | [link](https://cursor.com/dashboard/api) | Paste a Cursor user API key (crsr_...) from cursor.com/dashboard/api. OmniRoute exchanges it for a session token on demand; no IDE or cursor-agent install is needed. Usage bills to the Cursor plan that owns the key. |
+| `dahl` | `dahl` | Dahl | API key | [link](https://inference.dahl.global) | Click 'Add Account' to auto-generate a token, or add a manual API key. |
+| `databricks` | `databricks` | Databricks | API key, enterprise | [link](https://www.databricks.com) | — |
+| `datarobot` | `datarobot` | DataRobot | API key, enterprise | [link](https://docs.datarobot.com) | Use your DataRobot API token. Optional Base URL can be the account root (for LLM Gateway) or a deployment URL under /api/v2/deployments/. |
+| `deepai` | `deepai` | DeepAI | API key, image | [link](https://deepai.org) | Use your DeepAI API key. Get one at deepai.org — requires a Pro subscription ($9.99/mo). |
+| `deepinfra` | `deepinfra` | DeepInfra | API key | [link](https://deepinfra.com) | Free signup credits for API testing and model exploration |
+| `deepseek` | `ds` | DeepSeek | API key | [link](https://platform.deepseek.com) | 5M free tokens on signup - no credit card required |
+| `dgrid` | `dgrid` | DGrid | API key | [link](https://dgrid.ai) | DGrid Free Models Router: 10 requests/minute and 100 requests/day. A $5 lifetime top-up unlocks up to 20 requests/minute and 1,000 requests/day. |
+| `dify` | `dify` | Dify | API key | [link](https://dify.ai) | Get API key from your Dify instance. |
+| `digitalocean` | `digitalocean` | DigitalOcean | API key | [link](https://docs.digitalocean.com/products/ai-platform/) | — |
+| `dit` | `dai` | DIT.ai | API key | [link](https://dit.ai) | Use your dit.ai API key in Authorization: Bearer . Fully OpenAI-compatible — a drop-in replacement, just change the base URL to https://api.dit.ai/v1. |
+| `doubao` | `doubao` | Doubao | API key | [link](https://doubao.com) | Get API key at console.volcengine.com |
+| `dxnt` | `dxnt` | DXNT / DX Token | API key, aggregator | [link](https://www.dxnt.com) | Free accounts are documented at 100 calls/day; the quota may increase through invitations and can vary by account. |
+| `electronhub` | `electronhub` | Electron Hub | API key, aggregator | [link](https://www.electronhub.ai) | Free plan: 5 RPM, $0.25 weekly credits and 10 Neutrinos/day for :free models; family budgets also apply. |
+| `empower` | `empower` | Empower | API key, aggregator | [link](https://docs.empower.dev) | Bearer API key for the Empower OpenAI-compatible endpoint. |
+| `factory` | `factory` | Factory | API key | [link](https://factory.ai) | Bearer API key for the Factory OpenAI-compatible gateway. |
+| `fal-ai` | `fal` | Fal.ai | API key, image | [link](https://fal.ai) | — |
+| `fastrouter` | `fastrouter` | FastRouter | API key, aggregator | [link](https://fastrouter.ai) | Models with the :free suffix allow 10 requests/day per organization and model; availability may change. |
+| `featherless-ai` | `featherless` | Featherless AI | API key | [link](https://featherless.ai) | Free tier available — no credit card required |
+| `fenayai` | `fenayai` | FenayAI | API key, aggregator | [link](https://fenayai.com) | Bearer API key for the FenayAI OpenAI-compatible gateway. |
+| `fireworks` | `fireworks` | Fireworks AI | API key | [link](https://fireworks.ai) | $1 free starter credits on signup for API testing |
+| `free-ai` | `free-ai` | Free.ai | API key, aggregator | [link](https://free.ai) | 30,000 tokens/day cover self-hosted models after email verification. Usage beyond the pool can bill at raw cost, and premium external models are paid. |
+| `freeaiapikey` | `faik` | FreeAIAPIKey | API key | [link](https://freeaiapikey.com) | — |
+| `freebuff` | `freebuff` | Freebuff | API key | [link](https://freebuff.com) | Enter Freebuff / Codebuff Auth Token (obtained via CLI login or automated harvester). |
+| `freeinference` | `freeinference` | FreeInference | API key, aggregator | [link](https://freeinference.org) | Free research access without a card; non-Harvard applicants require manual approval and no numeric quota is publicly guaranteed. |
+| `freemodel-dev` | `fmd` | FreeModel.dev | API key | [link](https://freemodel.dev) | $300 free credits on signup — no credit card required. Access GPT-5.4 and GPT-5.5 (OpenAI's latest flagship models) through an OpenAI-compatible API. |
+| `freetheai` | `fta` | FreeTheAi | API key, aggregator | [link](https://freetheai.xyz) | Join the FreeTheAi Discord to get your free API key. |
+| `friendliai` | `friendli` | FriendliAI | API key | [link](https://friendli.ai) | Free tier for serverless inference — no credit card required |
+| `g4f-gemini` | `g4fgem` | g4f.space — Gemini | API key, aggregator | [link](https://g4f.space) | No auth required. Free tier is limited to 5 requests/minute — sign up at g4f.dev/members.html for higher limits. |
+| `g4f-groq` | `g4fgroq` | g4f.space — Groq | API key, aggregator | [link](https://g4f.space) | No auth required. Free tier is limited to 5 requests/minute — sign up at g4f.dev/members.html for higher limits. |
+| `g4f-nvidia` | `g4fnv` | g4f.space — NVIDIA | API key, aggregator | [link](https://g4f.space) | No auth required. Free tier is limited to 5 requests/minute — sign up at g4f.dev/members.html for higher limits. |
+| `g4f-ollama` | `g4foll` | g4f.space — Ollama | API key, aggregator | [link](https://g4f.space) | No auth required. Free tier is limited to 5 requests/minute — sign up at g4f.dev/members.html for higher limits. |
+| `g4f-pollinations` | `g4fpol` | g4f.space — Pollinations | API key, aggregator | [link](https://g4f.space) | No auth required. Free tier is limited to 5 requests/minute — sign up at g4f.dev/members.html for higher limits. |
+| `galadriel` | `galadriel` | Galadriel | API key | [link](https://galadriel.com) | ⚠️ **DEPRECATED.** api.galadriel.ai no longer resolves (sweep 2026-06-19); the inference API appears discontinued. |
+| `gemini` | `gemini` | Gemini (Google AI Studio) | API key | [link](https://aistudio.google.com) | Free tier available through Google AI Studio; current per-model quotas and regional limits apply |
+| `getgoapi` | `ggo` | GoAPI | API key, aggregator | [link](https://api.getgoapi.com) | — |
+| `gigachat` | `gigachat` | GigaChat (Sber) | API key | [link](https://developers.sber.ru) | — |
+| `gitlab` | `gitlab` | GitLab Duo PAT | API key | [link](https://docs.gitlab.com/user/duo_agent_platform/code_suggestions/) | GitLab personal access token for the public Code Suggestions API. Configure a self-hosted base URL when not using gitlab.com. |
+| `gitlawb` | `glb` | Gitlawb Opengateway (MiMo) | API key | [link](https://opengateway.gitlawb.com) | Free MiMo (xiaomi/mimo-v2.5) revoked 2026-05 — Opengateway is now a pay-as-you-go credit gateway; no recurring free model. |
+| `gitlawb-gmi` | `glb-gmi` | Gitlawb Opengateway (GMI Cloud) | API key | [link](https://opengateway.gitlawb.com) | Free Nemotron promo ended 2026-06 — the GMI Cloud route is now pay-as-you-go credit only. |
+| `glm` | `glm` | GLM Coding | API key | [link](https://z.ai/subscribe) | — |
+| `glm-cn` | `glmcn` | GLM Coding (China) | API key | [link](https://open.bigmodel.cn) | — |
+| `glmt` | `glmt` | GLM Thinking | API key | [link](https://open.bigmodel.cn) | — |
+| `groq` | `groq` | Groq | API key | [link](https://groq.com) | Free tier: 30 RPM / 14.4K RPD — no credit card |
+| `hackclub` | `hc` | Hackclub AI | API key, aggregator | [link](https://ai.hackclub.com) | Sign in with your Hack Club account at ai.hackclub.com. |
+| `haiper` | `hp` | Haiper | API key, video | [link](https://haiper.ai) | Get API key at haiper.ai/haiper-api |
+| `hcnsec` | `hcnsec` | Huancheng Public API | API key | [link](https://api.hcnsec.cn) | Get API key at api.hcnsec.cn |
+| `helixmind` | `helixmind` | HelixMind | API key, aggregator | [link](https://helixmind.online) | Previously circulated 3 RPM/50 RPD and no-card claims were not confirmed during the 2026-08-02 audit; current quota and billing require account verification. |
+| `helyxai` | `helyxai` | Helyx AI | API key, aggregator | [link](https://helyxai.space) | Operational Free plan documents 100,000 tokens/day; the site's separate 2M+ marketing claim conflicts and is not treated as a quota guarantee. |
+| `heroku` | `heroku` | Heroku AI | API key, enterprise | [link](https://www.heroku.com) | — |
+| `huggingface` | `hf` | HuggingFace | API key | [link](https://huggingface.co) | Free Inference API for thousands of models (Whisper, VITS, SDXL…) |
+| `hyperbolic` | `hyp` | Hyperbolic | API key | [link](https://hyperbolic.xyz) | $1-5 trial credits on signup for serverless inference |
+| `ideogram` | `ideo` | Ideogram | API key | [link](https://ideogram.ai) | Get API key at ideogram.ai/docs/api |
+| `iflytek` | `iflytek` | iFlytek Spark | API key | [link](https://xinghuo.xfyun.cn) | Get API key at console.xfyun.cn |
+| `inception` | `inception` | Inception | API key | [link](https://docs.inceptionlabs.ai) | 10M free tokens on signup, no credit card required. |
+| `inference-net` | `inet` | Inference.net | API key | [link](https://inference.net) | $25 free credits on signup plus research grants available |
+| `internlm` | `internlm` | InternLM (Intern-S1) | API key | [link](https://internlm.intern-ai.org.cn/) | Free monthly quota ~1M input / 3M output tokens (~10 RPM) |
+| `jina-ai` | `jina` | Jina AI (Foundation API) | API key, embed/rerank | [link](https://jina.ai) | Bearer API key for api.jina.ai — embeddings, rerank, classify, segment, and search. Dashboard keys take precedence over JINA_AI_API_KEY. This is not the Reader / r.jina.ai card and does not fetch URLs. |
+| `jina-reader` | `jr` | Jina Reader (r.jina.ai) | API key | [link](https://jina.ai/reader) | Bearer API key for r.jina.ai URL-to-markdown (/v1/web/fetch only). Does not serve /v1/embeddings or /v1/rerank. The same Jina token as Foundation API works; OmniRoute reuses a jina-ai dashboard key or JINA_AI_API_KEY when this card is empty. |
+| `kenari` | `kenari` | Kenari | API key | [link](https://kenari.id) | Use your Kenari API key (kn-...) in Authorization: Bearer . Fully OpenAI-compatible. API base URL: https://kenari.id/v1. |
+| `kie` | `kie` | KIE.AI | API key | [link](https://kie.ai) | — |
+| `kilo-gateway` | `kg` | Kilo Gateway | API key, aggregator | [link](https://kilo.ai) | — |
+| `kimi` | `kimi` | Kimi (Legacy Moonshot API) | API key | [link](https://platform.kimi.ai?aff=omniroute) | — |
+| `kimi-coding-apikey` | `kmca` | Kimi Code API Key | API key | [link](https://www.kimi.com/code?aff=omniroute) | — |
+| `lambda-ai` | `lambda` | Lambda AI | API key | [link](https://lambda.ai) | — |
+| `laozhang` | `lz` | LaoZhang AI | API key, aggregator | [link](https://api.laozhang.ai) | — |
+| `leonardo` | `leo` | Leonardo AI | API key, video | [link](https://leonardo.ai) | Get API key at leonardo.ai/developer |
+| `liquid` | `liquid` | Liquid AI | API key | [link](https://liquid.ai) | Get API key at liquid.ai |
+| `literouter` | `literouter` | LiteRouter | API key, aggregator | [link](https://literouter.com) | Free model variants use the :free suffix; daily credit limits vary by model and free input is capped at 5,000 tokens. |
+| `llamagate` | `llamagate` | LlamaGate | API key | [link](https://llamagate.ai) | — |
+| `llm-kiwi` | `llmkiwi` | LLM.Kiwi | API key, aggregator | [link](https://llm.kiwi) | Free plan exposes auto and hrLLM; the published 40 requests/hour limit applies to hrLLM. |
+| `llm7` | `llm7` | LLM7.io | API key | [link](https://llm7.io) | Use any non-empty key (for example 'unused'). If older built-in models return model_unavailable, use Available Models → Import from /models or Auto-Sync; verified live model: gemini-3.1-flash-lite. |
+| `llmgateway` | `llmgateway` | LLM Gateway | API key, aggregator | [link](https://llmgateway.io) | Hosted Free plan: free-priced models are limited to 5 requests per 10 minutes when the account has no credits. |
+| `logfare` | `logfare` | Logfare | API key, aggregator | [link](https://logfare.ai) | Create a free account at https://logfare.ai/register (username/password, no email verification) to get an instant API key, then paste it here as a Bearer token. |
+| `longcat` | `lc` | LongCat AI | API key | [link](https://longcat.chat/platform/docs) | Free: one-time 10M-token grant after account signup + KYC verification (LongCat-2.0). One-time only — not a recurring daily/monthly allowance. |
+| `magnific` | `freepik` | Magnific | API key, image | [link](https://www.magnific.com) | Get an API key at magnific.com/user/api-keys (header x-magnific-api-key). Legacy Freepik developer keys still work. |
+| `maritalk` | `maritalk` | Maritalk | API key | [link](https://www.maritaca.ai) | — |
+| `meganova-ai` | `meganova-ai` | MegaNova AI | API key, aggregator | [link](https://meganova.ai) | Free signup without a card. Published Tier 1 per-model quotas total 550 requests/day; they are not a shared global pool, and paid overage can apply if enabled. |
+| `meta-llama` | `meta` | Meta Llama API | API key | [link](https://llama.developer.meta.com) | — |
+| `minimax` | `minimax` | Minimax Coding | API key, video | [link](https://www.minimax.io) | — |
+| `minimax-cn` | `minimax-cn` | Minimax (China) | API key | [link](https://www.minimaxi.com) | — |
+| `mistral` | `mistral` | Mistral | API key | [link](https://mistral.ai) | Free Experiment tier: rate-limited access to all models, no credit card required |
+| `mixedbread` | `mxbai` | Mixedbread AI | API key | [link](https://www.mixedbread.com) | Bearer API key for the Mixedbread embeddings API. |
+| `mixlayer` | `mixlayer` | Mixlayer | API key, aggregator | [link](https://www.mixlayer.com) | The qwen/qwen3.5-4b-free model is free for prototyping and rate-limited; no fixed public RPM or daily quota is confirmed. |
+| `mnn-ai` | `mnn-ai` | MNN AI | API key, aggregator | [link](https://mnnai.ru) | Free plan: $1 monthly credits, 10 RPM and access only to models marked Free. |
+| `modal` | `mdl` | Modal | API key, enterprise | [link](https://modal.com/docs) | Use the bearer token that protects your Modal deployment, if enabled. Base URL should point to your OpenAI-compatible Modal app, for example https://--.modal.run/v1. |
+| `modelscope` | `ms` | ModelScope | API key | [link](https://modelscope.cn) | Free tier via ModelScope API-Inference — Alibaba account required. |
+| `monsterapi` | `monster` | MonsterAPI | API key | [link](https://monsterapi.ai) | ⚠️ **DEPRECATED.** Monster API shuttered operations on 2026-06-30. Use alternative OpenAI-compatible providers. |
+| `moonshot` | `moonshot` | Kimi | API key | [link](https://platform.kimi.ai?aff=omniroute) | — |
+| `morph` | `morph` | Morph | API key | [link](https://morphllm.com) | Free tier: 250K credits/month, $0 |
+| `muse-code` | `mc` | Muse Code (Meta) | API key | [link](https://github.com/meta-llama/llama-stack) | Use your META_API_KEY env var as a Bearer token. Muse Code CLI uses the OpenAI Responses API wire format (POST /responses). |
+| `naga-ac` | `naga` | Naga.ac | API key, aggregator | [link](https://naga.ac) | Get API key at naga.ac — Google/GitHub/Discord signup available. |
+| `naga-ai` | `naga-ai` | Naga AI | API key, aggregator | [link](https://naga.ac) | Models marked :free are publicly listed, but no numeric quota is confirmed. Naga's policy warns that free-tier prompts and outputs may be collected or used for training. |
+| `nanogpt` | `nanogpt` | NanoGPT | API key | [link](https://nano-gpt.com) | — |
+| `nara` | `nara` | NaraRouter | API key | [link](https://bynara.id) | Get a free API key via NaraRouter's Telegram channel, then paste it here as a Bearer token. |
+| `navy` | `navy` | NavyAI | API key | [link](https://api.navy) | Create a free API key from the NavyAI dashboard, then paste it here as a Bearer token. |
+| `nebius` | `nebius` | Nebius AI | API key | [link](https://nebius.com) | ~$1 trial credits on signup for API testing |
+| `nlpcloud` | `nlpc` | NLP Cloud | API key | [link](https://docs.nlpcloud.com) | Use your NLP Cloud API key in Authorization: Token . OmniRoute targets the chatbot endpoint on https://api.nlpcloud.io/v1/gpu//chatbot by default. |
+| `nomic` | `nomic` | Nomic | API key | [link](https://nomic.ai) | Get API key at atlas.nomic.ai |
+| `nous-research` | `nous` | Nous Research | API key | [link](https://portal.nousresearch.com/help) | Use your Nous Portal API key. OmniRoute targets the official OpenAI-compatible inference endpoint at https://inference-api.nousresearch.com/v1. |
+| `novita` | `novita` | Novita AI | API key, video, aggregator | [link](https://novita.ai) | $0.50 trial credits on signup (valid about 1 year) |
+| `nscale` | `nscale` | nScale | API key | [link](https://nscale.com) | $5 free credits on signup for inference testing |
+| `nube` | `nube` | Nube.sh | API key | [link](https://nube.sh) | — |
+| `nvidia` | `nvidia` | NVIDIA NIM | API key | [link](https://build.nvidia.com) | Free dev access: ~40 RPM, 70+ models (Kimi K2.5, GLM 4.7, DeepSeek V3.2...) |
+| `oci` | `oci` | OCI Generative AI | API key, enterprise | [link](https://www.oracle.com/artificial-intelligence/generative-ai) | Use your OCI Generative AI API key or IAM bearer token. Base URL can be https://inference.generativeai..oci.oraclecloud.com/openai/v1/. |
+| `ofoxai` | `ofoxai` | OfoxAI | API key, aggregator | [link](https://ofox.ai) | The current catalog advertises 10+ free models without a public numeric quota; review upstream provenance, retention and training terms before production use. |
+| `ollama-cloud` | `ollamacloud` | Ollama Cloud | API key | [link](https://ollama.com/settings/keys) | — |
+| `openadapter` | `oad` | OpenAdapter | API key | [link](https://openadapter.dev) | Use your OpenAdapter API key in Authorization: Bearer sk-cv-. Fully OpenAI-compatible. API base URL: https://api.openadapter.in/v1. |
+| `openai` | `openai` | OpenAI | API key | [link](https://platform.openai.com) | — |
+| `opencode-go` | `opencode-go` | OpenCode Go | API key | [link](https://opencode.ai/go) | — |
+| `opencode-zen` | `opencode-zen` | OpenCode Zen | API key | [link](https://opencode.ai/zen) | — |
+| `openference-api` | `ofa` | Openference API | API key | [link](https://openference.com) | Free plan: 3-day trial with open-source models — no credit card required |
+| `openrouter` | `openrouter` | OpenRouter | API key, aggregator | [link](https://openrouter.ai) | Free models at $0/token with :free suffix - 20 RPM / 200 RPD |
+| `openvecta` | `openvecta` | OpenVecta | API key | [link](https://openvecta.com) | Free credits on signup for OpenAI-compatible inference across LLMs, embeddings, and reasoning models |
+| `orcarouter` | `orcarouter` | OrcaRouter | API key | [link](https://www.orcarouter.ai) | — |
+| `ovhcloud` | `ovh` | OVHcloud AI | API key | [link](https://www.ovhcloud.com) | — |
+| `perplexity` | `pplx` | Perplexity | API key | [link](https://www.perplexity.ai) | — |
+| `piapi` | `pi` | PiAPI | API key, aggregator | [link](https://piapi.ai) | — |
+| `pioneer` | `pn` | Pioneer AI | API key | [link](https://pioneer.ai) | $75 free usage credits — no credit card required |
+| `plamo` | `plamo` | PLaMo | API key | [link](https://plamo.preferredai.jp/api) | — |
+| `poe` | `poe` | Poe | API key, aggregator | [link](https://creator.poe.com/api-reference) | Bearer API key for the Poe OpenAI-compatible API. |
+| `poixe-ai` | `poixe-ai` | Poixe AI | API key, aggregator | [link](https://poixe.com) | Current public free limits are small and model-group specific: 2 RPM/5 RPD for large-cup models and 20 RPM/50 RPD for small-cup models. |
+| `pollinations` | `pol` | Pollinations AI | API key, video | [link](https://pollinations.ai) | Anonymous/keyless access to the documented free models is best-effort. Local v3.8.50 verification (2026-07-31) returned 401 via OmniRoute and Cloudflare 1010 on direct upstream probes from the same network. Premium models still require a Pollinations API key from enter.pollinations.ai. |
+| `poolside` | `poolside` | Poolside | API key | [link](https://poolside.ai) | Laguna S 2.1 and XS 2.1 are free during Preview; no public numeric quota is published. |
+| `predibase` | `predibase` | Predibase | API key | [link](https://predibase.com) | ⚠️ **DEPRECATED.** serving.app.predibase.com no longer resolves (sweep 2026-06-19); the managed serving API appears discontinued. |
+| `publicai` | `publicai` | PublicAI | API key | [link](https://publicai.co) | Requires an API key — one-time signup credit, then paid |
+| `qianfan` | `qianfan` | Baidu Qianfan | API key | [link](https://cloud.baidu.com/product-s/qianfan_home) | — |
+| `qiniu` | `qiniu` | Qiniu | API key | [link](https://www.qiniu.com) | — |
+| `qwen-cloud` | `qwc` | Qwen Cloud | API key | [link](https://www.qwencloud.com/) | — |
+| `qwen-cloud-token-plan` | `qct` | Qwen Cloud Token Plan | API key | [link](https://www.qwencloud.com/pricing/token-plan) | — |
+| `recraft` | `recraft` | Recraft | API key, image | [link](https://recraft.ai) | — |
+| `regolo` | `regolo` | Regolo AI | API key | [link](https://regolo.ai) | Get your Regolo API key from regolo.ai, then paste it here as a Bearer token. |
+| `reka` | `reka` | Reka | API key | [link](https://docs.reka.ai/chat/overview) | Use your Reka API key. OmniRoute supports the OpenAI-compatible base URL https://api.reka.ai/v1 and sends both Authorization and X-Api-Key headers for compatibility. |
+| `requesty` | `requesty` | Requesty | API key | [link](https://requesty.ai) | Free tier ~200 requests/day - multi-model routing gateway (300+ models) |
+| `routeway` | `routeway` | Routeway | API key | [link](https://routeway.ai) | Create a free API key at routeway.ai, then paste it here as a Bearer token. |
+| `runwayml` | `runway` | Runway | API key, video | [link](https://docs.dev.runwayml.com) | Use your Runway API key in Authorization: Bearer . OmniRoute targets the current Runway API at https://api.dev.runwayml.com/v1 and sends the required X-Runway-Version header automatically. |
+| `sambanova` | `samba` | SambaNova | API key | [link](https://sambanova.ai) | $5 free credits on signup (30-day validity), no credit card required |
+| `sap` | `sap` | SAP Generative AI Hub | API key, enterprise | [link](https://help.sap.com/docs/sap-ai-core/sap-ai-core-service-guide/generative-ai-hub-in-sap-ai-core) | Use your SAP AI Core bearer token. Base URL can be your AI_API_URL root or a deploymentUrl from Generative AI Hub. |
+| `sarvam` | `sarvam` | Sarvam AI | API key | [link](https://docs.sarvam.ai) | ₹1,000 in free signup credits — never expire |
+| `scaleway` | `scw` | Scaleway AI | API key | [link](https://www.scaleway.com/en/docs/ai-data/generative-apis/) | 1M free tokens for new accounts — EU/GDPR compliant (Paris), Qwen3 235B & Llama 70B |
+| `sealion` | `sealion` | SEA-LION | API key | [link](https://sea-lion.ai) | Sign in at sea-lion.ai with Google (no card, no region wall), create an API key, then paste it here. |
+| `segmind` | `segmind` | Segmind | API key, image, video | [link](https://segmind.com) | Use your Segmind API key in the x-api-key header. OmniRoute targets https://api.segmind.com/v1/ and returns the generated image/video bytes directly. |
+| `sensenova` | `sensenova` | SenseNova | API key | [link](https://platform.sensenova.cn) | Get API key at platform.sensenova.cn |
+| `siliconflow` | `siliconflow` | SiliconFlow | API key | [link](https://cloud.siliconflow.com) | $1 free credits plus currently listed $0 models after identity verification; availability and limits may change |
+| `snowflake` | `snowflake` | Snowflake Cortex | API key, enterprise | [link](https://www.snowflake.com) | — |
+| `sparkdesk` | `sparkdesk` | SparkDesk | API key | [link](https://xinghuo.xfyun.cn) | Get API key at console.xfyun.cn |
+| `speka` | `speka` | Speka AI | API key, aggregator | [link](https://speka.me) | Free plan: $1 monthly usage, 10 RPM, one API key and access to open models and the playground; no card required. |
+| `stability-ai` | `stability` | Stability AI | API key, image | [link](https://stability.ai) | — |
+| `stepfun` | `stepfun` | StepFun | API key | [link](https://stepfun.com) | Get API key at platform.stepfun.com |
+| `sumopod` | `sumopod` | SumoPod | API key | [link](https://ai.sumopod.com) | Use your SumoPod API key (sk-...) in Authorization: Bearer . Fully OpenAI-compatible. API base URL: https://ai.sumopod.com/v1. |
+| `suno` | `suno` | Suno | API key | [link](https://suno.ai) | Paste session cookie from suno.ai (Clerk auth) |
+| `synthetic` | `synthetic` | Synthetic | API key, aggregator | [link](https://synthetic.new) | — |
+| `tabitoken` | `tabitoken` | TabiToken | API key, aggregator | [link](https://tabitoken.com) | — |
+| `tencent` | `tencent` | Tencent Hunyuan | API key | [link](https://hunyuan.tencent.com) | Get API key at console.cloud.tencent.com |
+| `thebai` | `thebai` | TheB.AI | API key, aggregator | [link](https://theb.ai) | Bearer API key for the TheB.AI OpenAI-compatible gateway. |
+| `tinyfish` | `tf` | TinyFish Fetch | API key | [link](https://docs.tinyfish.ai/fetch-api) | X-API-Key from agent.tinyfish.ai/api-keys |
+| `together` | `together` | Together AI | API key, video | [link](https://www.together.ai) | — |
+| `token-kiosk` | `tk` | Token Kiosk | API key | [link](https://agent-router.gaib.ai) | Use your Token Kiosk API key in Authorization: Bearer . Fully OpenAI-compatible gateway. API base URL: https://agent-router.gaib.ai/v1. |
+| `tokenreply` | `tokenreply` | TokenReply | API key, aggregator | [link](https://www.tokenreply.com) | Free-tagged models have model- and campaign-specific daily limits; no fixed global free quota is published. |
+| `tokenrouter` | `trk` | TokenRouter | API key | [link](https://tokenrouter.com) | Use your TokenRouter API key in Authorization: Bearer . Fully OpenAI-compatible. API base URL: https://api.tokenrouter.com/v1. |
+| `topaz` | `topaz` | Topaz | API key, image | [link](https://topazlabs.com) | — |
+| `typhoon` | `typhoon` | Typhoon | API key | [link](https://docs.opentyphoon.ai) | Free API key with a 5 req/s and 200 req/m rate limit. |
+| `udio` | `udio` | Udio | API key | [link](https://udio.com) | Paste session cookie from udio.com (Supabase auth) |
+| `uncloseai` | `unc` | UncloseAI | API key | [link](https://uncloseai.com) | No auth required. API accepts any non-empty string as key for identification. If older built-in models return 404, use Available Models → Import from /models or Auto-Sync; verified live model: solidrust/Hermes-3-Llama-3.1-8B-AWQ. |
+| `unorouter` | `unorouter` | UnoRouter | API key, aggregator | [link](https://unorouter.ai) | Models with the :free suffix do not debit balance; limit is 1 request/minute per free model per user. |
+| `upstage` | `upstage` | Upstage | API key | [link](https://www.upstage.ai) | — |
+| `v0-vercel` | `v0` | v0 (Vercel) | API key | [link](https://v0.dev) | — |
+| `venice` | `venice` | Venice.ai | API key | [link](https://venice.ai) | — |
+| `vercel-ai-gateway` | `vag` | Vercel AI Gateway | API key, aggregator | [link](https://vercel.com/docs/ai-gateway) | — |
+| `vertex` | `vertex` | Vertex AI | API key, enterprise | [link](https://cloud.google.com/vertex-ai) | Provide Service Account JSON or OAuth access_token |
+| `vertex-partner` | `vp` | Vertex AI Partners | API key, enterprise | [link](https://cloud.google.com/vertex-ai) | Provide the same Service Account JSON used for Vertex AI partner models. |
+| `void-ai` | `void-ai` | Void AI | API key, aggregator | [link](https://voidai.app) | The public model catalog marks some models with a free plan requirement, but access is conditional and no numeric quota is confirmed. |
+| `volcengine` | `volcengine` | Volcengine | API key | [link](https://www.volcengine.com) | — |
+| `voyage-ai` | `voyage` | Voyage AI | API key, embed/rerank | [link](https://www.voyageai.com) | Bearer API key for Voyage AI embeddings and rerank APIs. |
+| `wafer` | `wafer` | Wafer AI | API key | [link](https://wafer.ai) | — |
+| `wandb` | `wandb` | Weights & Biases Inference | API key | [link](https://wandb.ai) | — |
+| `watsonx` | `watsonx` | IBM watsonx.ai Gateway | API key, enterprise | [link](https://www.ibm.com/products/watsonx-ai) | Use your watsonx bearer token. Base URL can be https://.ml.cloud.ibm.com/ml/gateway/v1/ or a self-managed /ml/gateway/v1 endpoint. |
+| `writer` | `writer` | Writer | API key | [link](https://dev.writer.com) | — |
+| `x5lab` | `x5lab` | X5Lab | API key | [link](https://x5lab.dev) | Use your X5Lab API key (x5-...) in Authorization: Bearer . Fully OpenAI-compatible. API base URL: https://api.x5lab.dev/v1. |
+| `xai` | `xai` | xAI (Grok) | API key | [link](https://x.ai) | Use an official xAI API key, or sign in with xAI OAuth. Grok Build JWT sessions remain a separate provider. |
+| `xiaomi-mimo` | `mimo` | Xiaomi MiMo | API key | [link](https://mimo.mi.com) | — |
+| `xiaomi-mimo-token-plan` | `mimotp` | Xiaomi MiMo Token Plan | API key | [link](https://mimo.mi.com) | — |
+| `yi` | `yi` | Yi (01.AI) | API key | [link](https://01.ai) | Get API key at platform.lingyiwanwu.com |
+| `yolo-auto` | `yolo-auto` | Yolo-Auto | API key, aggregator | [link](https://yolo-auto.com) | Free API access is request-limited and intended for testing; no numeric daily quota is published and free access is not promised indefinitely. |
+| `zai` | `zai` | Z.AI | API key | [link](https://open.bigmodel.cn) | — |
+| `zenmux` | `zm` | ZenMux | API key | [link](https://zenmux.ai) | Use your ZenMux API key in Authorization: Bearer . ZenMux is fully OpenAI-compatible. Base URL: https://zenmux.ai/api/v1. |
+| `zerolimitai` | `zerolimitai` | ZeroLimitAI | API key, aggregator | [link](https://www.zerolimitai.com) | Temporary free trial is advertised, but official pages conflict between 3 and 7 days; a 100-calls/day claim is not treated as permanent. |
+| `zylo-api` | `zylo` | Zylo API | API key, aggregator | [link](https://zyloai.net) | Basic plan: 10 RPM, 7,200 requests/day and 200,000 tokens/day; limited to Basic text models. |
+
+## Local Providers (14)
+
+| ID | Alias | Name | Tags | Website | Notes |
+|----|-------|------|------|---------|-------|
+| `comfyui` | `comfyui` | ComfyUI | Local | [link](https://github.com/comfyanonymous/ComfyUI) | No API key required. Configure the local ComfyUI base URL (default: http://localhost:8188). |
+| `docker-model-runner` | `dmr` | Docker Model Runner | Local, self-hosted | [link](https://docs.docker.com/ai/model-runner/) | API key optional. Configure the local Docker Model Runner OpenAI-compatible base URL (default: http://localhost:12434/v1). |
+| `lemonade` | `lemonade` | Lemonade Server | Local, self-hosted | [link](https://lemonade-server.ai) | API key optional. Configure the local Lemonade OpenAI-compatible base URL (default: http://localhost:13305/api/v1). |
+| `llama-cpp` | `llamacpp` | llama.cpp | Local, self-hosted | [link](https://github.com/ggml-org/llama.cpp) | API key optional (use any value, e.g. sk-no-key-required). Configure the llama-server OpenAI-compatible base URL (default: http://127.0.0.1:8080/v1). Note: if Llamafile is also installed, both default to port 8080 — run only one at a time or override the port. |
+| `llamafile` | `llamafile` | Llamafile | Local, self-hosted | [link](https://github.com/Mozilla-Ocho/llamafile) | API key optional. Configure the local Llamafile OpenAI-compatible base URL (default: http://127.0.0.1:8080/v1). |
+| `lm-studio` | `lmstudio` | LM Studio | Local, self-hosted | [link](https://lmstudio.ai) | API key optional. Configure the local LM Studio OpenAI-compatible base URL (default: http://localhost:1234/v1). |
+| `mlx-gemma` | `mlx-gemma` | MLX Gemma 26B | Local, self-hosted | [link](https://github.com/ml-explore/mlx) | No API key required. Runs mlx-lm server locally on port 11435. Requires uv and mlx-lm installed. Model: mlx-community/gemma-4-26B-A4B-it-qat-q4_0-mlx-aligned (~15.9GB peak memory). |
+| `mlx-qwen` | `mlx-qwen` | MLX Qwen 3.8 27B | Local, self-hosted | [link](https://github.com/ml-explore/mlx) | No API key required. Runs mlx-lm server locally on port 11436. Requires uv and mlx-lm installed. Model: maglun/Qwen3.8-27B-MLX-Mixed-3.80bpw (~13.1GB peak memory). |
+| `ollama-local` | `ollama` | Ollama | Local, self-hosted | [link](https://ollama.com) | No API key required. Ollama runs locally — configure its OpenAI-compatible base URL (default: http://localhost:11434/v1) and make sure Ollama is running before connecting. |
+| `oobabooga` | `ooba` | oobabooga | Local, self-hosted | [link](https://github.com/oobabooga/text-generation-webui) | API key optional. Configure the local oobabooga OpenAI-compatible base URL (default: http://localhost:5000/v1). |
+| `sdwebui` | `sdwebui` | SD WebUI | Local | [link](https://github.com/AUTOMATIC1111/stable-diffusion-webui) | No API key required. Configure the local WebUI base URL (default: http://localhost:7860). |
+| `triton` | `triton` | NVIDIA Triton | Local, self-hosted | [link](https://developer.nvidia.com/triton-inference-server) | API key optional. Configure the Triton OpenAI-compatible base URL (default: http://localhost:8000/v1). |
+| `vllm` | `vllm` | vLLM | Local, self-hosted | [link](https://github.com/vllm-project/vllm) | API key optional. Configure the local vLLM OpenAI-compatible base URL (default: http://localhost:8000/v1). |
+| `xinference` | `xinference` | XInference | Local, self-hosted | [link](https://inference.readthedocs.io) | API key optional. Configure the local XInference OpenAI-compatible base URL (default: http://localhost:9997/v1). |
+
+## Search Providers (13)
+
+| ID | Alias | Name | Tags | Website | Notes |
+|----|-------|------|------|---------|-------|
+| `brave-search` | `brave-search` | Brave Search | Search | [link](https://brave.com/search/api) | Subscription token from Brave Search API dashboard |
+| `exa-search` | `exa-search` | Exa Search | Search | [link](https://exa.ai) | API key from dashboard.exa.ai |
+| `firecrawl` | `fc` | Firecrawl | Search | [link](https://firecrawl.dev) | API key from firecrawl.dev/app/api-keys (or set your self-hosted Firecrawl base URL) |
+| `google-pse-search` | `google-pse` | Google Programmable Search | Search | [link](https://developers.google.com/custom-search/v1/overview) | Requires a Google API key and your Programmable Search Engine ID (cx) |
+| `linkup-search` | `linkup` | Linkup Search | Search | [link](https://docs.linkup.so) | Bearer API key from the Linkup dashboard |
+| `ollama-search` | `ollama-search` | Ollama Search | Search | [link](https://ollama.com/settings/keys) | Same API key as Ollama Cloud (from ollama.com/settings/keys) |
+| `perplexity-search` | `pplx-search` | Perplexity Search | Search | [link](https://docs.perplexity.ai/guides/search-quickstart) | Same API key as Perplexity (pplx-...) |
+| `searchapi-search` | `searchapi` | SearchAPI | Search | [link](https://www.searchapi.io/docs/google) | API key from SearchAPI (query param or Bearer auth) |
+| `searxng-search` | `searxng` | SearXNG Search | Search | [link](https://docs.searxng.org) | API key is optional. Set your SearXNG base URL. Some instances may require a bearer token for access. |
+| `serper-search` | `serper-search` | Serper Search | Search | [link](https://serper.dev) | API key from serper.dev dashboard |
+| `tavily-search` | `tavily-search` | Tavily Search | Search | [link](https://tavily.com) | API key from app.tavily.com (format: tvly-...) |
+| `x-search` | `x_search` | X Search (Grok) | Search | [link](https://docs.x.ai/developers/tools/x-search) | SuperGrok OAuth (xai-oauth) or xAI API key. This is Grok X Search, not the X Developer MCP. |
+| `youcom-search` | `youcom-search` | You.com Search | Search | [link](https://you.com/business/api/) | X-API-Key from the You.com platform dashboard |
+
+## Audio-only Providers (12)
+
+| ID | Alias | Name | Tags | Website | Notes |
+|----|-------|------|------|---------|-------|
+| `assemblyai` | `aai` | AssemblyAI | Audio | [link](https://assemblyai.com) | — |
+| `aws-polly` | `polly` | AWS Polly | Audio | [link](https://aws.amazon.com/polly/) | Use AWS Secret Access Key as API key; set providerSpecificData.accessKeyId and optional region. |
+| `cartesia` | `cartesia` | Cartesia | Audio | [link](https://cartesia.ai) | — |
+| `deepgram` | `dg` | Deepgram | Audio | [link](https://deepgram.com) | — |
+| `elevenlabs` | `el` | ElevenLabs | Audio | [link](https://elevenlabs.io) | — |
+| `fishaudio` | `fishaudio` | Fish Audio | Audio | [link](https://fish.audio) | — |
+| `gladia` | `gladia` | Gladia | Audio | [link](https://gladia.io) | — |
+| `inworld` | `inworld` | Inworld | Audio | [link](https://inworld.ai) | — |
+| `playht` | `playht` | PlayHT | Audio | [link](https://play.ht) | — |
+| `rev-ai` | `revai` | Rev AI | Audio | [link](https://www.rev.ai) | — |
+| `soniox` | `sx` | Soniox | Audio | [link](https://soniox.com) | — |
+| `speechmatics` | `sm` | Speechmatics | Audio | [link](https://www.speechmatics.com) | Free tier — 8 hours/month, no credit card required. Batch (async) mode only. |
+
+## Upstream Proxy Providers (2)
+
+| ID | Alias | Name | Tags | Website | Notes |
+|----|-------|------|------|---------|-------|
+| `9router` | `nr` | 9router | Upstream proxy | [link](https://www.npmjs.com/package/9router) | — |
+| `cliproxyapi` | `cpa` | CLIProxyAPI | Upstream proxy | [link](https://github.com/router-for-me/CLIProxyAPI) | — |
+
+## Cloud Agent Providers (3)
+
+| ID | Alias | Name | Tags | Website | Notes |
+|----|-------|------|------|---------|-------|
+| `codex-cloud` | `codex-cloud` | Codex Cloud | Cloud agent | [link](https://openai.com/codex) | OpenAI API key with Codex Cloud task access. |
+| `devin` | `devin` | Devin | Cloud agent | [link](https://devin.ai) | Devin API key for cloud agent sessions. |
+| `jules` | `jules` | Google Jules | Cloud agent | [link](https://jules.google) | Jules API key for creating and managing cloud coding tasks. |
+
+## System Providers (1)
+
+| ID | Alias | Name | Tags | Website | Notes |
+|----|-------|------|------|---------|-------|
+| `auto` | `auto` | Auto (Zero-Config) | System | — | — |
+
+## Sources of truth
+
+- Catalog: [`src/shared/constants/providers.ts`](../../src/shared/constants/providers.ts)
+- Registry (per-model details): [`open-sse/config/providerRegistry.ts`](../../open-sse/config/providerRegistry.ts)
+- Executors: [`open-sse/executors/`](../../open-sse/executors/) (106 implementations)
+- Translators: [`open-sse/translator/`](../../open-sse/translator/)
+
+## See Also
+
+- [FREE_TIERS.md](./FREE_TIERS.md) — curated free-tier guide
+- [USER_GUIDE.md](../guides/USER_GUIDE.md) — provider setup walkthrough
+- [ARCHITECTURE.md](../architecture/ARCHITECTURE.md) — overall architecture
diff --git a/README.md b/README.md
index 55e6d0328a1..6600b406492 100644
--- a/README.md
+++ b/README.md
@@ -7,7 +7,7 @@
# 🚀 OmniRoute — The Free AI Gateway
-
+
@@ -101,7 +101,7 @@
@@ -461,7 +461,7 @@ All **19** strategies — mix & match per combo step:
-
+📊 Full methodology & per-feature detail vs 9router, OpenRouter, CLIProxyAPI & LiteLLM → [`docs/comparison/OMNIROUTE_VS_ALTERNATIVES.md`](docs/comparison/OMNIROUTE_VS_ALTERNATIVES.md)
@@ -559,7 +559,7 @@ the current catalog at **[radar.omniroute.online/planos](https://radar.omniroute
- **🖼️ New endpoints** — `/v1/ocr` (Mistral OCR) and `/v1/audio/translations` (Whisper-style) round out the media surface. → [API Reference](docs/reference/API_REFERENCE.md)
- **🎨 Image / video / audio generation** — one API for media: xAI Grok Imagine & Novita AI video, ComfyUI, Freepik, Adobe Firefly, Microsoft Designer, Segmind, EdgeTTS. → [API Reference](docs/reference/API_REFERENCE.md)
- **🌍 Deployment & ops** — reverse-proxy `basePath`, browser-language auto-detect, per-key device tracking, root-less MITM trust, zh-TW localization. → [Environment](docs/reference/ENVIRONMENT.md)
-- **🤝 More providers & agents** — Cursor Cloud Agent, Grok Build (xAI) with browser + OAuth login, Ollama first-class card, Claude Opus 5 & Sonnet 5, Kimi official partnership (Code/Web/Moonshot), Zed, Requesty, SenseNova, Yuanbao, Agnes AI… and a refreshed **348-provider catalog**. → [Providers](docs/reference/PROVIDER_REFERENCE.md)
+- **🤝 More providers & agents** — Cursor Cloud Agent, Grok Build (xAI) with browser + OAuth login, Ollama first-class card, Claude Opus 5 & Sonnet 5, Kimi official partnership (Code/Web/Moonshot), Zed, Requesty, SenseNova, Yuanbao, Agnes AI… and a refreshed **349-provider catalog**. → [Providers](docs/reference/PROVIDER_REFERENCE.md)
- **📡 Routing transparency** — every response carries an `X-OmniRoute-Decision` header naming the strategy/provider/latency that served it, a new `cache-optimized` combo strategy + Auto-Combo `cacheAffinity` factor route repeat requests back to the connection holding the cached prefix, and a read-only `/v1/auto-combo/{channel}/candidates` endpoint exposes an `auto/*` channel's live candidate pool. → [Auto-Combo](docs/routing/AUTO-COMBO.md)
- **⚡ Local performance & infra** — one-click local Redis, Cloudflare Workers / Deno Deploy relay deployers, Bifrost & Mux as supervised embedded services. → [Embedded Services](docs/frameworks/EMBEDDED-SERVICES.md)
@@ -642,11 +642,11 @@ of your shell history. → [CLI Integrations](docs/guides/CLI-INTEGRATIONS.md)
-## 🌐 348 AI Providers — 90+ Free
+## 🌐 349 AI Providers — 90+ Free
-> The most complete catalog of any open-source router: **348 providers**, **90+ with a free tier**, **56 free forever**.
+> The most complete catalog of any open-source router: **349 providers**, **90+ with a free tier**, **56 free forever**.
@@ -990,11 +990,11 @@ docker run -d --name omniroute --restart unless-stopped --stop-timeout 40 \
`:latest` follows the highest **published** stable SemVer. It does not track git `main`. Pin `:X.Y.Z` for GitOps. See [Docker Release Channels](docs/guides/DOCKER_GUIDE.md#release-channels).The image pins **`OMNIROUTE_MEMORY_MB=1024`**. That is enough for the dashboard and a light chat. **Coding agents** (`POST /v1/responses` from Claude Code, Codex, Grok, …) need a much larger V8 heap or the process `FATAL ERROR`s at ~12 GiB under two overlapping long contexts. Size the container above the heap (native buffers sit outside V8):
-| Workload | Heap (`-e OMNIROUTE_MEMORY_MB`) | Container (`--memory`) |
-| --- | --- | --- |
-| Dashboard / light chat | `1024` (image default) | ≥2 g |
-| One coding agent | `8192` | ≥10 g |
-| Two concurrent long `/v1/responses` | `10240`–`12288` | ≥12–16 g |
+| Workload | Heap (`-e OMNIROUTE_MEMORY_MB`) | Container (`--memory`) |
+| ----------------------------------- | ------------------------------- | ---------------------- |
+| Dashboard / light chat | `1024` (image default) | ≥2 g |
+| One coding agent | `8192` | ≥10 g |
+| Two concurrent long `/v1/responses` | `10240`–`12288` | ≥12–16 g |
```bash
docker run -d --name omniroute --restart unless-stopped --stop-timeout 40 \
@@ -1003,6 +1003,7 @@ docker run -d --name omniroute --restart unless-stopped --stop-timeout 40 \
```
Full table: [Docker Guide — runtime RAM](docs/guides/DOCKER_GUIDE.md#runtime-ram-for-coding-agents).
+
> **Pre-release Docker channel:** `diegosouzapw/omniroute:next` and
> `diegosouzapw/omniroute:next-web` follow the current default `release/v*`
> branch. These mutable tags are intended only for testing unreleased fixes and
diff --git a/bin/cli/commands/combo.mjs b/bin/cli/commands/combo.mjs
index 554c1c1d838..8d58cf73bd9 100644
--- a/bin/cli/commands/combo.mjs
+++ b/bin/cli/commands/combo.mjs
@@ -307,6 +307,12 @@ export async function runComboCreateCommand(name, strategy = "priority", opts =
}
const models = Array.isArray(opts.models) ? opts.models : [];
+ if (!models.length) {
+ console.error(
+ "combo create requires at least one target. Pass --models and/or repeat --model ."
+ );
+ return 1;
+ }
try {
return await withRuntime(async ({ kind, api, db }) => {
diff --git a/changelog.d/features/10987-logfare-free-provider.md b/changelog.d/features/10987-logfare-free-provider.md
new file mode 100644
index 00000000000..507a528411f
--- /dev/null
+++ b/changelog.d/features/10987-logfare-free-provider.md
@@ -0,0 +1 @@
+- **feat(providers):** add Logfare as a free OpenAI-compatible provider — dashboard card with a Free badge and request-logging disclosure (every prompt/completion is logged for research; opt out at logfare.ai/consent), live model discovery from `https://logfare.ai/v1/models` (20 models, 11 chat-capable: kimi-k3, deepseek-v4-pro, glm-5.2, gpt-5.6-luna, minimax-m3…), full chat/streaming through the existing OpenAI-compatible path, the real Logfare logo on the card, and a listing in the free-tiers guide. ([#10987](https://github.com/diegosouzapw/OmniRoute/pull/10987))
diff --git a/changelog.d/features/11190-usage-command-json.md b/changelog.d/features/11190-usage-command-json.md
new file mode 100644
index 00000000000..d7655f04c51
--- /dev/null
+++ b/changelog.d/features/11190-usage-command-json.md
@@ -0,0 +1 @@
+- **feat(api):** `/api/usage/om-usage` gains a structured form — `?format=json` returns the key's own usage as `ApiKeyUsageLimitStatus` + `UsageSnapshot` instead of `text/plain`. This is the surface a UI (the OmniCopilot panel) consumes to show a key holder their daily/weekly spend and quota reset. The route is self-service (the caller's own key, gated by `allowUsageCommand`), not the management surface; refusals come back as a discriminated `{ "allowed": false, "error": … }` so a UI can tell "not allowed" apart from "allowed but nothing cached yet". The endpoint was previously undocumented in `API_REFERENCE.md`; it now has a section ([#11190](https://github.com/diegosouzapw/OmniRoute/pull/11190))
diff --git a/changelog.d/features/11192-usage-command-providers-array.md b/changelog.d/features/11192-usage-command-providers-array.md
new file mode 100644
index 00000000000..b7ef4211091
--- /dev/null
+++ b/changelog.d/features/11192-usage-command-providers-array.md
@@ -0,0 +1 @@
+- **feat(api):** `/api/usage/om-usage?format=json` now returns `providers[]` — every connection's quota snapshot, not just the single selected one — so a panel can render Codex / Claude / OpenCode side by side. The collector already gathered all of them; the single-pick `provider` field (kept) is a terminal presentation choice. Closes the per-connection gap from OmniCopilot #8 ([#11192](https://github.com/diegosouzapw/OmniRoute/pull/11192))
diff --git a/changelog.d/fixes/10071-g4f-space-anonymous-tier-proof-of-work.md b/changelog.d/fixes/10071-g4f-space-anonymous-tier-proof-of-work.md
new file mode 100644
index 00000000000..47f8de84fd2
--- /dev/null
+++ b/changelog.d/fixes/10071-g4f-space-anonymous-tier-proof-of-work.md
@@ -0,0 +1 @@
+- **fix(providers):** the five g4f.space sub-providers (Groq, Gemini, Pollinations, Ollama, NVIDIA) no longer advertise a free tier — a keyless `POST /v1/chat/completions` now returns `402 insufficient_credits` behind a proof-of-work "cake" wall (re-verified live 2026-08-22), so `hasFree` is `false` and the notes point at `g4f.dev/members.html`. The gateway still works with a member key, so its registry wiring and `authType: "optional"` are unchanged ([#10071](https://github.com/diegosouzapw/OmniRoute/issues/10071)) — thanks @chirag127
diff --git a/changelog.d/fixes/10550-responses-reasoning-transport.md b/changelog.d/fixes/10550-responses-reasoning-transport.md
index d34c433debd..e2b40cdb8c9 100644
--- a/changelog.d/fixes/10550-responses-reasoning-transport.md
+++ b/changelog.d/fixes/10550-responses-reasoning-transport.md
@@ -1 +1 @@
-- Preserve portable plaintext reasoning by default across streaming and non-streaming Chat Completions and Responses routes while keeping provider-bound opaque state target-compatible. Combos now drop incompatible continuation reasoning by default and can explicitly skip incompatible targets, while known providers no longer show redundant encrypted-reasoning controls. (#10550)
+- Preserve portable plaintext reasoning by default across streaming and non-streaming Chat Completions and Responses routes while keeping provider-bound opaque state target-compatible. Direct requests drop incompatible continuation reasoning by default; combos can explicitly skip incompatible targets without mutating the request. Known providers no longer show redundant encrypted-reasoning controls. (#10550, #10959)
diff --git a/changelog.d/fixes/10949-mixed-reasoning-plaintext.md b/changelog.d/fixes/10949-mixed-reasoning-plaintext.md
new file mode 100644
index 00000000000..05a055ec538
--- /dev/null
+++ b/changelog.d/fixes/10949-mixed-reasoning-plaintext.md
@@ -0,0 +1 @@
+- Preserve explicit plaintext reasoning when a Responses reasoning item also carries opaque provider state (rare OpenCode Go `deepseek-v4-flash` responses). Mixed plaintext + opaque input is projected onto the target transport: plaintext targets keep portable text, opaque targets keep provider state. Opaque-only reasoning is dropped when the selected target cannot replay it, allowing cross-model conversations to continue. (#10949, #10959)
diff --git a/changelog.d/fixes/11015-shutdown-track-sse.md b/changelog.d/fixes/11015-shutdown-track-sse.md
new file mode 100644
index 00000000000..1ed99b3669d
--- /dev/null
+++ b/changelog.d/fixes/11015-shutdown-track-sse.md
@@ -0,0 +1 @@
+- **fix(resilience):** count heavyweight `/v1` admission leases in the SIGTERM drain and send `Retry-After` on shutdown 503s so Recreate no longer looks like an empty 502 ([#11015](https://github.com/diegosouzapw/OmniRoute/issues/11015)) — thanks @RaviTharuma
diff --git a/changelog.d/fixes/11089-chat-routing-synced-inventory.md b/changelog.d/fixes/11089-chat-routing-synced-inventory.md
new file mode 100644
index 00000000000..922b96a659b
--- /dev/null
+++ b/changelog.d/fixes/11089-chat-routing-synced-inventory.md
@@ -0,0 +1 @@
+- **fix(resilience):** filter chat connection selection by each connection's *synced* model inventory on multi-host self-hosted providers (`ollama-local`, `lm-studio`, `vllm`, …), so a request for a model only one host advertises is pinned to that host instead of failing over onto a host that never had it ([#11089](https://github.com/diegosouzapw/OmniRoute/issues/11089))
diff --git a/changelog.d/fixes/11149-opencode-go-flat-rate.md b/changelog.d/fixes/11149-opencode-go-flat-rate.md
new file mode 100644
index 00000000000..7aa63ad4559
--- /dev/null
+++ b/changelog.d/fixes/11149-opencode-go-flat-rate.md
@@ -0,0 +1 @@
+- **fix(analytics):** `opencode-go` is now classified as a flat-rate subscription, so cost analytics shows $0 for it instead of billing every call at the underlying model’s metered rate — it resells GLM, Kimi, Grok, DeepSeek, MiniMax, Qwen and GPT-5.x under one flat monthly fee, which made the overstatement large rather than marginal ([#11149](https://github.com/diegosouzapw/OmniRoute/pull/11149)) — thanks @electrumguy
diff --git a/changelog.d/fixes/11162-combo-create-requires-model.md b/changelog.d/fixes/11162-combo-create-requires-model.md
new file mode 100644
index 00000000000..228e6a9b20f
--- /dev/null
+++ b/changelog.d/fixes/11162-combo-create-requires-model.md
@@ -0,0 +1 @@
+- **Combo create:** creating a routing combo without any model is now refused (`400`) — the CLI requires `--models`/`--model` on `combo create`, matching the dashboard which already rejected empty combos.
diff --git a/changelog.d/fixes/11180-keyless-custom-provider-auto-pool.md b/changelog.d/fixes/11180-keyless-custom-provider-auto-pool.md
new file mode 100644
index 00000000000..c533feae543
--- /dev/null
+++ b/changelog.d/fixes/11180-keyless-custom-provider-auto-pool.md
@@ -0,0 +1 @@
+- **fix(routing):** a custom `openai-compatible-*` / `anthropic-compatible-*` connection pointing at a keyless self-hosted backend (llama.cpp, Ollama, vLLM started without an API key) now stays in the `auto/*` candidate pool instead of being silently dropped by the credential gate — for those IDs "no credential" is the normal configuration, not an unconfigured connection ([#11180](https://github.com/diegosouzapw/OmniRoute/pull/11180)) — thanks @marcs7
diff --git a/changelog.d/fixes/11181-lkgp-enabled-context.md b/changelog.d/fixes/11181-lkgp-enabled-context.md
new file mode 100644
index 00000000000..d1c0cde5a36
--- /dev/null
+++ b/changelog.d/fixes/11181-lkgp-enabled-context.md
@@ -0,0 +1 @@
+- **fix(routing):** the Routing tab's "last known good provider" toggle now actually takes effect — `lkgpEnabled` was persisted and the `lkgp` strategy guarded on it, but the setting was never forwarded into the `RoutingContext` built in `resolveAutoStrategyOrder()`, so `context.lkgpEnabled` was always `undefined` and the off-switch was unreachable ([#11181](https://github.com/diegosouzapw/OmniRoute/issues/11181))
diff --git a/changelog.d/fixes/9763-ratelimit-mintime-floor.md b/changelog.d/fixes/9763-ratelimit-mintime-floor.md
new file mode 100644
index 00000000000..2b145f1e1a3
--- /dev/null
+++ b/changelog.d/fixes/9763-ratelimit-mintime-floor.md
@@ -0,0 +1 @@
+- **fix(ratelimit):** respect operator `minTimeBetweenRequestsMs` floor when relaxing the limiter on headroom — the adaptive rate-limit learning no longer silently erases a configured minimum gap between requests when the upstream reports plenty of remaining capacity ([#9763](https://github.com/diegosouzapw/OmniRoute/issues/9763)).
diff --git a/changelog.d/fixes/pending-opencode-empty-rejection-rotation.md b/changelog.d/fixes/pending-opencode-empty-rejection-rotation.md
new file mode 100644
index 00000000000..82a88c5905a
--- /dev/null
+++ b/changelog.d/fixes/pending-opencode-empty-rejection-rotation.md
@@ -0,0 +1 @@
+- **fix(executors):** OpencodeExecutor rotates (or retries once on a single-account direct path) on upstream 400 empty-body rejections — malformed completion envelopes with no error field were propagated as success and killed client sessions. Bounded +1 attempt per request; body reads are conditioned on status 400 so successful/streaming responses are never buffered. 400s carrying an error field keep propagating immediately.
diff --git a/config/quality/eslint-suppressions.json b/config/quality/eslint-suppressions.json
index 3dae8591dd7..79875e31489 100644
--- a/config/quality/eslint-suppressions.json
+++ b/config/quality/eslint-suppressions.json
@@ -853,11 +853,6 @@
"count": 1
}
},
- "src/app/api/usage/call-logs/route.ts": {
- "no-restricted-imports": {
- "count": 1
- }
- },
"src/app/api/usage/quota/route.ts": {
"no-restricted-imports": {
"count": 1
@@ -953,11 +948,6 @@
"count": 1
}
},
- "src/app/api/v1/rerank/route.ts": {
- "no-restricted-imports": {
- "count": 1
- }
- },
"src/app/api/v1/vscode/[token]/models/route.ts": {
"no-restricted-syntax": {
"count": 1
@@ -3259,4 +3249,4 @@
"count": 5
}
}
-}
+}
\ No newline at end of file
diff --git a/config/quality/file-size-baseline.json b/config/quality/file-size-baseline.json
index b7f40cebfa6..0a3239ea7fb 100644
--- a/config/quality/file-size-baseline.json
+++ b/config/quality/file-size-baseline.json
@@ -1,5 +1,6 @@
{
"_rebaseline_2026_08_20_10531_freebuff_provider": "PR #10531 (adrianaryaputra, feat/freebuff-provider-support, closes #6793) own growth: src/shared/constants/providers/apikey/gateways.ts 1283->1298 (+15, the freebuff APIKEY_PROVIDERS_GATEWAYS catalog entry, additive data at the existing registry chokepoint, same god-file no-split rationale as prior gateways.ts rebaselines) and src/app/(dashboard)/dashboard/providers/[id]/components/modals/AddApiKeyModal.tsx 1062->1067 (+5, freebuff credential placeholder/hint at the existing per-provider switch chokepoint). Covered by tests/unit/freebuff-provider.test.ts (9/9 passing).",
+ "_rebaseline_2026_08_21_10987_logfare_provider": "PR #10987 (jonlwheat2-gif, feat/10644-logfare-provider, closes #10644) own growth: src/shared/constants/providers/apikey/gateways.ts 1298->1321 (+23, the logfare APIKEY_PROVIDERS_GATEWAYS catalog entry with Free badge/freeNote/apiHint documenting the request-logging policy, additive data at the existing registry chokepoint, same god-file no-split rationale as the prior gateways.ts rebaselines: #10531 freebuff, merge-storm 2026-08-11). Covered by tests/unit/logfare-registry.test.ts (1/1 passing).",
"_rebaseline_2026_08_20_10574_reasoning_transport_fallback": "PR #10574 (jackjinke, fix/responses-reasoning-transport, fixes #10550) own growth: src/sse/handlers/chatHelpers.ts 1017->1019 (+2 = the new reasoningTransportFallback option threaded through executeChatWithBreaker's options destructure and its downstream handleSingleModel call, at the existing per-attempt options-passthrough chokepoint; not extractable without splitting the option-forwarding call itself). Covered by the PR's own reasoning-policy test suite (tests/unit/chatcore-translation-paths.test.ts, tests/unit/combo-attempt-body-isolation-7847.test.ts, tests/unit/reasoning-cache.test.ts, tests/unit/strip-reasoning-blobs-agentic-context-1599.test.ts among others), 446/446 focused tests passing.",
"_rebaseline_2026_08_18_10517_zed_hosted_oauth_callback_port": "PR #10517 (phatchau036, fix/zed-hosted-oauth-callback-port) own growth: src/shared/components/OAuthModal.tsx 1131->1148 (wc -l; check-file-size.mjs counts via split(\"\\n\").length so the gate sees 1134->1149, +15/+18, crosses the frozen 1134 cap). Wires the zed-hosted native-app callback auto-complete: forceManual gating on isTrueLocalhost for zed-hosted, the loopback-redirect-URI comment block, and the exchangeToken full-URL-as-code branch, all at the existing provider-switch chokepoints this modal already carries growth for (seventh bump: 969->989->993->998->1030->1056->1100->1149; structural shrink tracked in #3501). The actual port-derivation logic lives in src/lib/oauth/providers/zed-hosted.ts (not frozen here) and was hardened during pre-merge review to use the server's own getRuntimePorts() instead of a browser-guessed scheme/port, covered by the new tests/unit/zed-hosted-loopback-port-derivation.test.ts (8/8 passing).",
"_rebaseline_2026_08_13_10243_codex_fingerprint_merge": "PR #10243 (xz-dev, Codex OAuth fingerprint convergence) merge into release/v3.8.50: src/app/(dashboard)/dashboard/providers/[id]/providerPageHelpers.ts crossed the 1000-line new-file cap for the first time (974 on base, 997 on the PR's own branch, 1013 after merging + prettier reflow) purely from combining two independent, already-legitimate feature additions that landed on the same shared UI-helper file — this PR's own Codex fingerprint-mode select/toggle wiring (CODEX_FINGERPRINT_MODE_VALUES, getCodexFingerprintModeLabel, CodexFingerprintModeValue) plus #8949's unrelated Codex account-service-tier helpers merged concurrently on release/v3.8.50. Neither addition alone crosses the cap; git's line-level auto-merge does not detect a threshold crossing. Not modularized as part of this conflict-resolution merge commit (out of scope — this is a merge, not a feature change). Covered by the PR's own tests/unit/codex-fingerprint-convergence.test.ts, tests/unit/executor-codex.test.ts, tests/unit/provider-specific-data-schema.test.ts (all passing post-merge).",
@@ -388,6 +389,10 @@
"open-sse/services/claudeCodeCompatible.ts": 1563,
"open-sse/services/combo.ts": 4742,
"open-sse/services/compression/strategySelector.ts": 1379,
+ "open-sse/services/compression/engines/ccr/index.ts": 1024,
+ "_rebaseline_2026_08_22_11084_ccr_caller_gate": "PR #11084 (HouMinXi) own growth: open-sse/services/compression/engines/ccr/index.ts 1000->1024 (first listing — the engine was unlisted and drifted just over the 1000 cap; +24 are the callerSupportsCcrRetrieve gate that skips replacement entirely for callers without the retrieve tool, closing the stranded-prompt incident measured in production). Covered by tests/unit/compression/ccr-non-mcp-full-prompt-loss-7746.test.ts. Owner pre-authorized baseline bumps 2026-08-22.",
+ "open-sse/services/contextManager.ts": 1001,
+ "_rebaseline_2026_08_22_11113_purify_system_first": "PR #11113 (ggdayup) own growth: open-sse/services/contextManager.ts 1000->1001 (+1, purifyHistory merges the compression notice into the leading system message instead of splicing a second one mid-array — live-confirmed TokenRouter 400s; the +1 is the merge-into-leading branch, not extractable). Covered by tests/unit/context-manager-purify-system-first.test.ts. Owner pre-authorized baseline bumps 2026-08-22.",
"open-sse/services/rateLimitManager.ts": 1517,
"open-sse/translator/response/openai-responses.ts": 1652,
"open-sse/utils/cursorAgentProtobuf.ts": 1956,
@@ -427,7 +432,8 @@
"src/shared/components/analytics/charts.tsx": 1346,
"src/shared/services/cliRuntime.ts": 1459,
"src/sse/handlers/chat.ts": 2493,
- "src/sse/services/auth.ts": 3260,
+ "src/sse/services/auth.ts": 3337,
+ "_rebaseline_2026_08_23_11186_synced_inventory_routing": "PR #11186 (pacocartones) own growth: src/sse/services/auth.ts 3260->3337 (+77, loadAdvertisedModelsForSelfHostedConnections + the modelNotAdvertised candidate-filter predicate — pins chat routing to the connection whose synced inventory actually advertises the model, fixing spurious model-not-found on multi-host self-hosted setups; at the existing credential-selection chokepoint, not extractable without splitting the selection flow). Covered by tests/unit/chat-routing-synced-inventory-11089.test.ts. Owner pre-authorized baseline bumps 2026-08-22.",
"tests/unit/account-fallback-service.test.ts": 2044,
"tests/unit/provider-validation-specialty.test.ts": 3880,
"open-sse/executors/hyperagent.ts": 1334,
@@ -436,7 +442,8 @@
"open-sse/executors/kiro.ts": 1390,
"open-sse/translator/request/openai-to-kiro.ts": 1374,
"open-sse/utils/sseHeartbeat.ts": 194,
- "open-sse/utils/proxyFetch.ts": 1239,
+ "open-sse/utils/proxyFetch.ts": 1244,
+ "_rebaseline_2026_08_23_11177_dns_retry_classification": "PR #11177 (rqzbeh) own growth: proxyFetch.ts 1239->1244 (+5, EAI_AGAIN/ENOTFOUND/ETIMEDOUT join the retryable dispatcher classification alongside ECONNREFUSED — bounded socket retries for transient DNS failures, part of the #10443 Hermes→Antigravity stream-drop fixes). Covered by tests/unit/proxy-fetch-dns-retry-10443.test.ts. Owner pre-authorized baseline bumps 2026-08-22.",
"_rebaseline_2026_08_11_v3850_merge_storm_provider_registry: DRIFT do merge-storm 2026-08-11 (99 PRs mergeados no release/v3.8.50). AddApiKeyModal.tsx (PR #8949 ChatGPT Web provider) e useProviderConnections.ts/ModelSelectModal.tsx (PRs #9011 combo test-all, #9499 image combos) = UI nova legítima acima do cap; gateways.ts = god-file de catálogo de providers que cresceu com os PRs #9009/#9421/#9468/#9594 (qualquer split arriscaria corromper o merge de novo — o próprio PR #9421 foi o que quebrou o arquivo; sem split até o release, congelado no tamanho atual). Owner autorizou rebaseline com anotação (2026-08-11).": {
"src/app/(dashboard)/dashboard/providers/[id]/components/modals/AddApiKeyModal.tsx": 1062,
"src/app/(dashboard)/dashboard/providers/[id]/hooks/useProviderConnections.ts": 1051,
@@ -447,7 +454,7 @@
"_rebaseline_2026_08_22_11156_enter_check_disabled": "PR #11156 (rqzbeh) own growth: AddApiKeyModal.tsx 1080->1082 (+2, Enter keydown handler now mirrors the isCheckDisabled condition — owner-requested post-merge polish from #11056; the rest of the diff is Prettier reflow). Covered by tests/unit/ui/add-api-key-modal-enter-key.test.tsx (jsdom render test, Enter dispatch assertions).",
"src/app/(dashboard)/dashboard/providers/[id]/hooks/useProviderConnections.ts": 1051,
"src/shared/components/ModelSelectModal.tsx": 1138,
- "src/shared/constants/providers/apikey/gateways.ts": 1298,
+ "src/shared/constants/providers/apikey/gateways.ts": 1321,
"open-sse/vendor/codex-chatgpt-web/bridge.ts": 1387,
"_rebaseline_2026_08_11_v3850_merge_storm_provider_registry": "DRIFT do merge-storm 2026-08-11 (99 PRs mergeados no release/v3.8.50). AddApiKeyModal.tsx (PR #8949 ChatGPT Web provider) e useProviderConnections.ts/ModelSelectModal.tsx (PRs #9011 combo test-all, #9499 image combos) = UI nova legitima acima do cap; gateways.ts = god-file de catalogo de providers que cresceu com PRs #9009/#9421/#9468/#9594 (qualquer split arriscaria corromper o merge de novo — o proprio PR #9421 quebrou o arquivo); bridge.ts (PR #8949) = ponte Chromium vendored; proxyFetch.ts 1207->1220 = drift herdado de merges. Owner autorizou rebaseline com anotacao (2026-08-11).",
"src/lib/modelCapabilities.ts": 1072,
@@ -455,7 +462,8 @@
"src/app/(dashboard)/dashboard/providers/[id]/providerPageHelpers.ts": 1014,
"open-sse/config/imageRegistry.ts": 1034,
"src/sse/handlers/chatHelpers.ts": 1019,
- "src/shared/middleware/chatBodyAdmission.ts": 1005,
+ "src/shared/middleware/chatBodyAdmission.ts": 1009,
+ "_rebaseline_2026_08_22_11020_sigterm_drain": "PR #11020 (RaviTharuma) own growth: chatBodyAdmission.ts 1005->1009 (+4, heavyweight admission leases now increment the SIGTERM drain counter and releaseChatAdmissionWhenDone holds it for the SSE lifetime — closes #11015; +4 are the lease/drain wiring lines at the existing admission chokepoint). Covered by tests/unit/chat-body-admission.test.ts heavyweight-lease cases. Owner pre-authorized baseline bumps 2026-08-22.",
"_rebaseline_2026_08_20_10668_tabitoken_gateway": "#10668 (yawar-aquil) own catalog growth: src/shared/constants/providers/apikey/gateways.ts 1268->1283 (+15, entirely this PR diff -- one new tabitoken gateway entry, data lines only; base moved from 1255 to 1268 via other merges since the PR forked). Not combination drift: reproducible on the PR branch alone, so the WS5.5 release-captain rule does not apply. Extraction is not available -- the file is pure data (own header: \"Pure data; merged by apikey/index.ts via spread\") and already split into 6 family files under apikey/. Same precedent as _rebaseline_2026_08_14_imagetotext_servicekinds (#10275/#10291, gateways.ts 1250->1255, data lines only) and _rebaseline_2026_08_11_v3850_merge_storm_provider_registry (owner-authorized for this same file).",
"open-sse/executors/commandCode.ts": 1059,
"_rebaseline_2026_08_21_10859_vision_bridge_catalog": "#10859 own growth (Vision Bridge fixes #10808/#10809): src/lib/modelCapabilities.ts 1006->1016 (+10, cmd/gpt-5.3-codex* text-only capability resolution) and open-sse/executors/commandCode.ts 988->1023 (+35, Command Code wire-model normalization for bare ids + reasoning field fallback for opencode-routed gateways). Cohesive bug fixes at the existing capability-resolution / executor chokepoints; not extractable mid-fix. Covered by tests/unit/model-capabilities-command-code-codex-textonly-10703.test.ts, tests/unit/command-code-vision.test.ts, tests/unit/opencode-mimo-reasoning-details-nonstream.test.ts. Pushed directly to release (own-session miss: the original rebaseline was made in a throwaway validation worktree and never landed on the PR branch or the release before merge).",
diff --git a/config/quality/quality-baseline.json b/config/quality/quality-baseline.json
index 95e35a5adec..604a5a8e19d 100644
--- a/config/quality/quality-baseline.json
+++ b/config/quality/quality-baseline.json
@@ -197,10 +197,11 @@
"_rebaseline_2026_08_09_v3850_release_close": "7666 -> 8045 (+379 gzip bytes, +4.9%). Release v3.8.50 close reconciliation measured twice with the real size-limit + @size-limit/file path on tip e0ce95c592. Per-entry measurements remain below their absolute budgets: omniroute.mjs 4380/15000, mcp-server.mjs 1195/5000, nodeRuntimeSupport.mjs 887/8000, reset-password.mjs 1583/6000. The growth accumulated through legitimate CLI/runtime work in this cycle, including global-install ESM alias resolution, Termux cache preparation, and MCP stdio startup hardening; no entrypoint is near its absolute ceiling. The direction:down ratchet stays blocking from this exact measured tip."
},
"openapiBreaking": {
- "value": 0,
+ "value": 4,
"direction": "down",
"dedicatedGate": true,
- "_note": "oasdiff breaking-change gate (Fase 9 Onda 0). Blocks any breaking change vs base spec."
+ "_note": "oasdiff breaking-change gate (Fase 9 Onda 0). Blocks any breaking change vs base spec.",
+ "_rebaseline_2026_08_22_combo_create_min1": "0 -> 4, split 3 own + 1 inherited. Docs-only alignment of components.schemas.ComboCreate with the request contract already enforced by the API since 638fc5fbd (combo create refuses an empty model list) and d5034ea52: `model`/`nodes` were phantom properties the server never accepted, and `models` (array, minItems 1) is the real required field. OWN findings (3, caused by this commit): removed `model`, removed `nodes`, added required `models` on POST /api/combos — spec-vs-server drift, not client-facing breakage, no working client could have relied on the removed shapes. INHERITED finding (1, NOT caused by this PR's code changes — pre-existing drift already present at parent d5034ea52): PATCH /api/combos/{id} request-body-added-required; that route's patch operation declares its own inline requestBody (required: true, bare object schema, docs/openapi.yaml ~2107-2118) and does not reference ComboCreate, so this finding exists independently of the ComboCreate alignment (same own-growth vs inherited-drift convention as _rebaseline_2026_07_20_aliasresolver_hook_split_7808). No code change in this PR; follow-up tracking = this change's PR description."
},
"mutationScore.src/sse/services/auth.ts": {
"value": 52.57,
diff --git a/docs/changelog/fragments/10962.md b/docs/changelog/fragments/10962.md
new file mode 100644
index 00000000000..5170a415e3f
--- /dev/null
+++ b/docs/changelog/fragments/10962.md
@@ -0,0 +1 @@
+fix(catalog): expose only provider-routable GLM reasoning-effort tiers and remove unroutable ZCode aliases
diff --git a/docs/diagrams/cli-terminal.svg b/docs/diagrams/cli-terminal.svg
index e2ad57c8b13..6ec68793d0c 100644
--- a/docs/diagrams/cli-terminal.svg
+++ b/docs/diagrams/cli-terminal.svg
@@ -1,4 +1,4 @@
-
import("./components/AddCompatibleProviderModal"),
+ { ssr: false }
+);
import { CategoryDot } from "./components/CategoryDot";
-import { ImportProvidersFromFileModal } from "./components/ImportProvidersFromFileModal";
+const ImportProvidersFromFileModal = dynamic(
+ () =>
+ import("./components/ImportProvidersFromFileModal").then(
+ (m) => m.ImportProvidersFromFileModal
+ ),
+ { ssr: false }
+);
import NoAuthProvidersSection from "./components/NoAuthProvidersSection";
import HighlightableProviderCard from "./components/HighlightableProviderCard";
import ProviderCountBadge from "./components/ProviderCountBadge";
diff --git a/src/app/(dashboard)/dashboard/usage/components/ProviderLimits/utils.tsx b/src/app/(dashboard)/dashboard/usage/components/ProviderLimits/utils.tsx
index 3af32da285d..dc102936d09 100644
--- a/src/app/(dashboard)/dashboard/usage/components/ProviderLimits/utils.tsx
+++ b/src/app/(dashboard)/dashboard/usage/components/ProviderLimits/utils.tsx
@@ -98,7 +98,7 @@ export function formatQuotaLabel(name: string) {
return `Weekly ${toTitleCaseWords(weeklyModelMatch[1])}`;
}
- return trimmed;
+ return toTitleCaseWords(trimmed.replace(/_/g, " "));
}
/**
diff --git a/src/app/(dashboard)/home/page.tsx b/src/app/(dashboard)/home/page.tsx
index beccde22610..bc10df88f42 100644
--- a/src/app/(dashboard)/home/page.tsx
+++ b/src/app/(dashboard)/home/page.tsx
@@ -4,6 +4,7 @@ import { getSettings } from "@/lib/localDb";
import HomePageClient from "../dashboard/HomePageClient";
import BootstrapBanner from "../dashboard/BootstrapBanner";
import KimiSponsorBanner from "../dashboard/KimiSponsorBanner";
+import CheaperInferenceSponsorBanner from "../dashboard/CheaperInferenceSponsorBanner";
import VscodeCopilotBanner from "../dashboard/VscodeCopilotBanner";
import NewsBanner from "../dashboard/NewsBanner";
@@ -20,6 +21,7 @@ export default async function HomePage() {
<>
{isBootstrapped && }
+
diff --git a/src/app/api/providers/[id]/models/discovery/providerSets.ts b/src/app/api/providers/[id]/models/discovery/providerSets.ts
index 82386814790..2b35e54b01e 100644
--- a/src/app/api/providers/[id]/models/discovery/providerSets.ts
+++ b/src/app/api/providers/[id]/models/discovery/providerSets.ts
@@ -95,6 +95,11 @@ export const NAMED_OPENAI_STYLE_PROVIDERS = new Set([
"internlm",
"ant-ling",
"nanogpt",
+ // Logfare (https://logfare.ai) — free OpenAI-compatible gateway live-verified
+ // 2026-08-21: GET https://logfare.ai/v1/models returns a real 20-model catalog
+ // (11 chat-capable). Live fetch keeps it fresh; the registry seed stays as the
+ // offline fallback.
+ "logfare",
]);
export function isNamedOpenAIStyleProvider(provider: string): boolean {
diff --git a/src/app/api/providers/[id]/models/route.ts b/src/app/api/providers/[id]/models/route.ts
index 01f1bd72857..cecbb6776eb 100755
--- a/src/app/api/providers/[id]/models/route.ts
+++ b/src/app/api/providers/[id]/models/route.ts
@@ -1909,12 +1909,9 @@ export async function GET(
}
if (isAnthropicCompatibleProvider(provider)) {
- const cachedResponse = maybeReturnCachedDiscovery();
- if (cachedResponse) return cachedResponse;
-
- const autoFetchDisabledResponse = maybeReturnAutoFetchDisabled();
- if (autoFetchDisabledResponse) return autoFetchDisabledResponse;
-
+ // CC providers never support models listing — this check must precede
+ // the cached-discovery / auto-fetch fallbacks, which would otherwise
+ // return a misleading 200 "no models" for a CC node (#10828 ordering).
if (isClaudeCodeCompatibleProvider(provider)) {
return NextResponse.json(
{ error: `Provider ${provider} does not support models listing` },
@@ -1922,6 +1919,12 @@ export async function GET(
);
}
+ const cachedResponse = maybeReturnCachedDiscovery();
+ if (cachedResponse) return cachedResponse;
+
+ const autoFetchDisabledResponse = maybeReturnAutoFetchDisabled();
+ if (autoFetchDisabledResponse) return autoFetchDisabledResponse;
+
let baseUrl = getProviderBaseUrl(connection.providerSpecificData);
if (!baseUrl) {
const fallback = buildDiscoveryFallbackResponse({
diff --git a/src/app/login/page.tsx b/src/app/login/page.tsx
index f2c70306f38..664eb3c1721 100644
--- a/src/app/login/page.tsx
+++ b/src/app/login/page.tsx
@@ -38,8 +38,7 @@ export default function LoginPage() {
if (data.nodeVersion) setNodeVersion(data.nodeVersion);
if (data.nodeCompatible === false) setNodeCompatible(false);
if (data.authenticated === true || data.requireLogin === false) {
- router.push("/dashboard");
- router.refresh();
+ window.location.href = "/dashboard";
return;
}
setHasPassword(!!data.hasPassword);
@@ -77,13 +76,12 @@ export default function LoginPage() {
if (res.ok) {
sessionStorage.setItem("omniroute_login_time", String(Date.now()));
- router.push("/dashboard");
- router.refresh();
+ window.location.href = "/dashboard";
} else {
const data = await res.json();
// (#521) If no password is set, redirect to onboarding instead of showing an error
if (data.needsSetup) {
- router.push("/dashboard/onboarding");
+ window.location.href = "/dashboard/onboarding";
return;
}
setError(data.error || t("invalidPassword"));
diff --git a/src/domain/connectionModelRules.ts b/src/domain/connectionModelRules.ts
index 316ade72d80..7831bbc84b6 100644
--- a/src/domain/connectionModelRules.ts
+++ b/src/domain/connectionModelRules.ts
@@ -80,3 +80,26 @@ export function hasEligibleConnectionForModel(
(connection) => !isModelExcludedByConnection(modelId, connection?.providerSpecificData)
);
}
+
+/**
+ * #11089: does this connection's *synced* inventory advertise the model?
+ *
+ * Unlike `excludedModels` (a manually maintained denylist) this reads the
+ * per-connection catalog written by model discovery, so a multi-host local
+ * provider never routes a model to a host that never had it. Ids are matched
+ * with the same candidate semantics as the denylist (provider prefix and the
+ * `[1m]` extended-context suffix are tolerated), but never as wildcard
+ * patterns — a synced id is a literal.
+ *
+ * Fails OPEN on an empty inventory: a host that has not been synced yet is
+ * "unknown", not "does not have it".
+ */
+export function isModelAdvertisedByConnection(
+ modelId: unknown,
+ advertisedModelIds: ReadonlySet | null | undefined
+): boolean {
+ if (!advertisedModelIds || advertisedModelIds.size === 0) return true;
+ if (typeof modelId !== "string" || modelId.trim().length === 0) return true;
+
+ return getModelMatchCandidates(modelId).some((candidate) => advertisedModelIds.has(candidate));
+}
diff --git a/src/i18n/messages/en.json b/src/i18n/messages/en.json
index ff778753c61..930f5b59587 100644
--- a/src/i18n/messages/en.json
+++ b/src/i18n/messages/en.json
@@ -13869,5 +13869,12 @@
"toolsMismatch": "Provider does not support tool calling",
"structuredOutputMismatch": "Provider does not support structured output",
"contextWindowMismatch": "Request exceeds provider context window"
+ },
+ "cheaperInferenceSponsorBanner": {
+ "title": "Cheaper Inference is an OmniRoute Open Source Friend",
+ "description": "A cost-ranked gateway reselling dozens of frontier models behind one OpenAI-compatible endpoint — routing each request to the cheapest eligible provider, never above list price.",
+ "cta": "Get an API Key",
+ "partnerLinkNote": "Partner link",
+ "dismissAriaLabel": "Dismiss"
}
}
diff --git a/src/i18n/messages/pt-BR.json b/src/i18n/messages/pt-BR.json
index f3e3c15d130..8df20f0f9df 100644
--- a/src/i18n/messages/pt-BR.json
+++ b/src/i18n/messages/pt-BR.json
@@ -4922,7 +4922,9 @@
"multiProvider": "Multi-Provedor",
"usageTracking": "Rastreamento de Uso",
"securityDesc": "Defina uma senha para proteger seu painel, ou pule por enquanto.",
+ "securityDescSkipWarning": "⚠️ Sem uma senha, você não poderá adicionar provedores durante a configuração. Você poderá adicioná-los depois pelo painel, após definir uma senha.",
"providerDesc": "Conecte seu primeiro provedor de IA. Você pode adicionar mais depois.",
+ "providerRequiresPassword": "Você precisa definir uma senha primeiro para adicionar provedores. Volte à etapa de segurança e defina uma senha, ou adicione provedores depois pelo painel.",
"apiKeyRequired": "Chave de API (obrigatório)",
"customUrlOptional": "URL personalizada (opcional)",
"testDesc": "Vamos verificar se a conexão com seu provedor funciona.",
@@ -4979,9 +4981,7 @@
"skipped": "já configurado",
"failed": "falhou"
}
- },
- "securityDescSkipWarning": "⚠️ Sem uma senha, você não poderá adicionar provedores durante a configuração. Você pode adicioná-los depois no painel após definir uma senha.",
- "providerRequiresPassword": "Você precisa definir uma senha primeiro para adicionar provedores. Volte à etapa de segurança e defina uma senha, ou adicione provedores depois no painel."
+ }
},
"providers": {
"title": "Provedores",
@@ -6269,6 +6269,20 @@
"webSessionGuideStep3": "Copie a credencial necessária do próprio domínio do provedor. Para cookies, copie apenas o valor do cabeçalho Cookie e omita Cookie:.",
"webSessionGuideStep3Manual": "Caminho manual: abra as ferramentas do desenvolvedor do navegador (F12 → Network), atualize a página, abra uma requisição autenticada e copie o valor do cabeçalho Cookie em Request Headers — omita o prefixo Cookie:.",
"webSessionGuideStep4": "Cole aqui e verifique a conexão. Se parar de funcionar, faça login novamente e substitua-o por um novo valor.",
+ "harImportButtonLabel": "Importar arquivo .har",
+ "harImportButtonBusy": "Importando…",
+ "harImportButtonHint": "Exporte pela aba Rede das Ferramentas do Desenvolvedor após enviar pelo menos uma mensagem no chat.",
+ "harImportStatusValid": "Importado — válido por cerca de {minutes} min.",
+ "harImportStatusExpiringSoon": "Importado — válido por apenas mais cerca de {minutes} min.",
+ "harImportStatusExpired": "Importado, mas este token expirou há {minutes} min — exporte um HAR novo.",
+ "harImportStatusUnknownExpiry": "Importado. Não foi possível ler a expiração.",
+ "harImportErrorNotJson": "Esse arquivo não é um JSON válido — ele é realmente uma exportação .har?",
+ "harImportErrorNoEntries": "Este HAR não contém entradas de rede.",
+ "harImportErrorNoChathubUrl": "Nenhuma conexão de chat do Copilot foi encontrada neste HAR. Envie pelo menos uma mensagem em m365.cloud.microsoft antes de exportar.",
+ "harImportErrorUnparsableUrl": "A conexão de chat foi encontrada, mas não foi possível ler a URL.",
+ "harImportErrorMissingFields": "A conexão de chat foi encontrada, mas o token estava ausente.",
+ "harImportErrorReadFailed": "Não foi possível ler esse arquivo.",
+ "harImportErrorUnknown": "Não foi possível extrair uma credencial desse arquivo HAR.",
"webSessionSecurityHint": "Trate isso como uma senha: ela poderá acessar sua conta da web conectada até que ela expire ou seja revogada.",
"webNoAuthGuideTitle": "Nenhuma credencial necessária",
"webNoAuthGuideBody": "{provider} não precisa de chave de API ou cookie. Salve a conexão para usar seu endpoint web gratuito.",
diff --git a/src/i18n/messages/pt.json b/src/i18n/messages/pt.json
index 11a3606a54c..1921f3fba7f 100644
--- a/src/i18n/messages/pt.json
+++ b/src/i18n/messages/pt.json
@@ -13845,5 +13845,12 @@
"toolsMismatch": "Provider does not support tool calling",
"structuredOutputMismatch": "Provider does not support structured output",
"contextWindowMismatch": "Request exceeds provider context window"
+ },
+ "cheaperInferenceSponsorBanner": {
+ "title": "A Cheaper Inference é uma Amiga do Código Aberto do OmniRoute",
+ "description": "Um gateway com custo ordenado que revende dezenas de modelos de fronteira num único endpoint compatível com OpenAI — roteando cada requisição ao provedor elegível mais barato, nunca acima do preço de tabela.",
+ "cta": "Obter uma Chave de API",
+ "partnerLinkNote": "Link de parceiro",
+ "dismissAriaLabel": "Dispensar"
}
}
diff --git a/src/lib/memory/injection.ts b/src/lib/memory/injection.ts
index 9fdd2bbbf10..d4d8ead7f7c 100644
--- a/src/lib/memory/injection.ts
+++ b/src/lib/memory/injection.ts
@@ -65,7 +65,11 @@ export function providerSupportsSystemMessage(provider: string | null | undefine
*
* Populated with the Xiaomi MiMo endpoint (provider id `xiaomi-mimo`, registry
* alias `mimo`, serving mimo-v2.5) confirmed live to 400 on a non-first system
- * message. Add other providers here only when they are documented as strict.
+ * message, and the TokenRouter gateway (provider id `tokenrouter`), confirmed
+ * live on 2026-08-22 to reject mid-array system messages — including the
+ * compression notice spliced by purifyHistory() before that splice was fixed to
+ * merge into the leading system message. Add other providers here only when
+ * they are documented as strict.
*
* Self-hosted deployments can extend this list without a source change via
* OMNIROUTE_STRICT_SYSTEM_PROVIDERS (comma-separated provider ids,
@@ -73,7 +77,7 @@ export function providerSupportsSystemMessage(provider: string | null | undefine
* self-hosted Qwen3.5+/3.6 model, whose chat template enforces the same
* single-leading-system-message constraint as xiaomi-mimo.
*/
-const BUILTIN_PROVIDERS_SYSTEM_MUST_BE_FIRST = new Set(["xiaomi-mimo", "mimo"]);
+const BUILTIN_PROVIDERS_SYSTEM_MUST_BE_FIRST = new Set(["xiaomi-mimo", "mimo", "tokenrouter"]);
/**
* Parses OMNIROUTE_STRICT_SYSTEM_PROVIDERS into a normalized id list.
diff --git a/src/lib/modelMetadataRegistry.ts b/src/lib/modelMetadataRegistry.ts
index 3828047b90b..b7aa2aa2164 100644
--- a/src/lib/modelMetadataRegistry.ts
+++ b/src/lib/modelMetadataRegistry.ts
@@ -130,6 +130,11 @@ function uniqueStrings(values: Array) {
];
}
+export function isGlmFamilyModel(modelId: string, displayName = ""): boolean {
+ const glmFamilyPattern = /(?:^|[/@:_. -])glm(?=$|[-._ /@:](?:z)?\d|\d)/i;
+ return glmFamilyPattern.test(modelId) || glmFamilyPattern.test(displayName);
+}
+
function toQualifiedId(
providerAlias: string | null,
provider: string | null,
@@ -477,11 +482,16 @@ export function enrichCatalogModelEntry(
? declaredEffortTiers
: sourceDeclaresThinking
? undefined
- : extendCodexGpt56EffortValues(
- metadata.provider,
- metadata.model,
- CANONICAL_EFFORT_VALUES
- );
+ : // #10963: GLM-family models never inherit generic OpenAI tiers — an
+ // explicit empty list is authoritative unless a provider-declared
+ // contract exists (handled by declaredEffortTiers above).
+ isGlmFamilyModel(metadata.model, metadata.displayName)
+ ? []
+ : extendCodexGpt56EffortValues(
+ metadata.provider,
+ metadata.model,
+ CANONICAL_EFFORT_VALUES
+ );
const capabilityFields = {
...(typeof metadata.capabilities.vision === "boolean"
? { vision: metadata.capabilities.vision }
@@ -502,7 +512,9 @@ export function enrichCatalogModelEntry(
// #6241: surface thinking support + the canonical effort tiers so the frontend can
// render the effort/thinking toggles. `thinking` is kept for back-compat; `supportsThinking`
// is the explicit flag and `effort_tiers` lists the selectable reasoning levels
- // (only when the model actually supports thinking).
+ // (only when the model actually supports thinking). An explicit empty registry list
+ // is authoritative; GLM models also require a provider-declared contract instead of
+ // inheriting generic OpenAI effort tiers.
...(typeof metadata.capabilities.supportsThinking === "boolean"
? {
thinking: metadata.capabilities.supportsThinking,
diff --git a/src/lib/proxyLogger.ts b/src/lib/proxyLogger.ts
index a9eb4b3805d..8665e20c758 100644
--- a/src/lib/proxyLogger.ts
+++ b/src/lib/proxyLogger.ts
@@ -194,43 +194,103 @@ export function logProxyEvent(entry: ProxyLogInput) {
proxyLogs.length = MAX_IN_MEMORY_ENTRIES;
}
- // 2. Persist to SQLite
+ // 2. Queue for background batch persistence (SQLite / Redis)
if (shouldPersistToDisk) {
+ enqueueProxyLog(log);
+ }
+
+ return log;
+}
+
+// ──────────────── Background Batch Persistence ────────────────
+
+const BATCH_FLUSH_INTERVAL_MS = 1000;
+const BATCH_SIZE_THRESHOLD = 100;
+
+let pendingLogsQueue: ProxyLogEntry[] = [];
+let batchTimer: NodeJS.Timeout | null = null;
+
+function ensureBatchTimer() {
+ if (batchTimer) return;
+ batchTimer = setInterval(() => {
+ flushProxyLogsSync();
+ }, BATCH_FLUSH_INTERVAL_MS);
+ if (typeof batchTimer.unref === "function") {
+ batchTimer.unref();
+ }
+}
+
+function enqueueProxyLog(log: ProxyLogEntry) {
+ pendingLogsQueue.push(log);
+ ensureBatchTimer();
+ if (pendingLogsQueue.length >= BATCH_SIZE_THRESHOLD) {
+ flushProxyLogsSync();
+ }
+}
+
+export function flushProxyLogsSync() {
+ if (pendingLogsQueue.length === 0) return;
+ const batch = pendingLogsQueue;
+ pendingLogsQueue = [];
+
+ // 1. If Redis driver is active, asynchronously publish batch to Redis Stream/Channel
+ if (process.env.QUOTA_STORE_DRIVER === "redis" || process.env.QUOTA_STORE_REDIS_URL) {
try {
- const db = getDbInstance();
- db.prepare(
- `INSERT INTO proxy_logs (id, timestamp, status, proxy_type, proxy_host, proxy_port,
- level, level_id, provider, target_url, public_ip, egress_ip, latency_ms, error,
- connection_id, combo_id, account, tls_fingerprint)
- VALUES (@id, @timestamp, @status, @proxyType, @proxyHost, @proxyPort,
- @level, @levelId, @provider, @targetUrl, @clientIp, @egressIp, @latencyMs, @error,
- @connectionId, @comboId, @account, @tlsFingerprint)`
- ).run({
- id: log.id,
- timestamp: log.timestamp,
- status: log.status,
- proxyType: log.proxy?.type || null,
- proxyHost: log.proxy?.host || null,
- proxyPort: log.proxy?.port ? Number(log.proxy.port) : null,
- level: log.level,
- levelId: log.levelId,
- provider: log.provider,
- targetUrl: log.targetUrl,
- clientIp: log.clientIp,
- egressIp: log.egressIp,
- latencyMs: log.latencyMs,
- error: log.error,
- connectionId: log.connectionId,
- comboId: log.comboId,
- account: log.account,
- tlsFingerprint: log.tlsFingerprint ? 1 : 0,
- });
- } catch (err: any) {
- console.warn("[proxyLogger] Failed to persist:", err.message);
+ import("@/lib/quota/redisQuotaStore").then(({ getRedisQuotaStore }) => {
+ const store = getRedisQuotaStore(process.env.QUOTA_STORE_REDIS_URL || "");
+ const client = (store as any)?.client;
+ if (client && typeof client.publish === "function") {
+ for (const entry of batch) {
+ client.publish("omniroute:proxy_logs", JSON.stringify(entry)).catch(() => {});
+ }
+ }
+ }).catch(() => {});
+ } catch {
+ /* ignore redis pub errors */
}
}
- return log;
+ // 2. Persist to SQLite using a single transaction for high-performance non-blocking write
+ try {
+ const db = getDbInstance();
+ const insertStmt = db.prepare(
+ `INSERT INTO proxy_logs (id, timestamp, status, proxy_type, proxy_host, proxy_port,
+ level, level_id, provider, target_url, public_ip, egress_ip, latency_ms, error,
+ connection_id, combo_id, account, tls_fingerprint)
+ VALUES (@id, @timestamp, @status, @proxyType, @proxyHost, @proxyPort,
+ @level, @levelId, @provider, @targetUrl, @clientIp, @egressIp, @latencyMs, @error,
+ @connectionId, @comboId, @account, @tlsFingerprint)`
+ );
+
+ const transaction = db.transaction((entries: ProxyLogEntry[]) => {
+ for (const item of entries) {
+ insertStmt.run({
+ id: item.id,
+ timestamp: item.timestamp,
+ status: item.status,
+ proxyType: item.proxy?.type || null,
+ proxyHost: item.proxy?.host || null,
+ proxyPort: item.proxy?.port ? Number(item.proxy.port) : null,
+ level: item.level,
+ levelId: item.levelId,
+ provider: item.provider,
+ targetUrl: item.targetUrl,
+ clientIp: item.clientIp,
+ egressIp: item.egressIp,
+ latencyMs: item.latencyMs,
+ error: item.error,
+ connectionId: item.connectionId,
+ comboId: item.comboId,
+ account: item.account,
+ tlsFingerprint: item.tlsFingerprint ? 1 : 0,
+ });
+ }
+ });
+
+ transaction(batch);
+ } catch (err: any) {
+ console.warn("[proxyLogger] Failed to write proxy log batch to disk:", err?.message || err);
+ }
}
// ──────────────── Query ────────────────
diff --git a/src/lib/services/portProbe.ts b/src/lib/services/portProbe.ts
index 2a9fd502bda..9a890f541b4 100644
--- a/src/lib/services/portProbe.ts
+++ b/src/lib/services/portProbe.ts
@@ -168,11 +168,24 @@ export function parseSsPid(stdout: string): number | null {
export function parseNetstatPid(stdout: string, port: number): number | null {
for (const line of stdout.split("\n")) {
const columns = line.trim().split(/\s+/);
- // proto recv-q send-q local-address foreign-address state pid/program
+ // Linux: proto recv-q send-q local-address foreign-address state pid/program
if (columns.length < 7 || columns[5] !== "LISTEN") continue;
- if (!columns[3].endsWith(`:${port}`)) continue;
- const parsed = Number.parseInt(columns[6], 10);
- if (Number.isFinite(parsed)) return parsed;
+ const linuxAddress = columns[3].endsWith(`:${port}`);
+ const macAddress = columns[3].endsWith(`.${port}`);
+ if (!linuxAddress && !macAddress) continue;
+
+ if (linuxAddress) {
+ const linuxPid = Number.parseInt(columns[6], 10);
+ if (Number.isFinite(linuxPid)) return linuxPid;
+ }
+
+ // macOS `netstat -anv -p tcp` appends a `process:pid` column after
+ // the socket counters. Process names may contain spaces, so scan instead
+ // of relying on one fixed column index.
+ for (const column of columns.slice(6)) {
+ const match = /:(\d+)$/.exec(column);
+ if (match) return Number.parseInt(match[1], 10);
+ }
}
return null;
}
@@ -197,7 +210,11 @@ const PID_PROBES: ReadonlyArray<{
args: (port) => ["-tlnp", `sport = :${port}`],
parse: (stdout) => parseSsPid(stdout),
},
- { command: "netstat", args: () => ["-tlnp"], parse: parseNetstatPid },
+ {
+ command: "netstat",
+ args: () => (process.platform === "darwin" ? ["-anv", "-p", "tcp"] : ["-tlnp"]),
+ parse: parseNetstatPid,
+ },
];
/** Run one probe, resolving null on a missing binary, a non-match or a timeout. */
diff --git a/src/lib/usage/flatRateProviders.ts b/src/lib/usage/flatRateProviders.ts
index 8d3eec2c5d7..3454b9b3917 100644
--- a/src/lib/usage/flatRateProviders.ts
+++ b/src/lib/usage/flatRateProviders.ts
@@ -46,6 +46,11 @@ const FLAT_RATE_SUBSCRIPTION_PROVIDER_IDS: ReadonlySet = new Set([
"glm-cn", // GLM Coding (China) plan
"claude", // Claude Code plan (OAuth-only — a Claude Pro/Max subscription)
"cc", // Claude Code plan (alias id — same connection, shares the `cc` pricing rows)
+ // OpenCode Go subscription (https://opencode.ai/go) — a flat monthly fee. It is an
+ // aggregator reselling GLM, Kimi, Grok, DeepSeek, MiniMax, Qwen and GPT-5.x, so
+ // per-token rows price each call at the UNDERLYING model's metered rate and the
+ // analytics overstatement is large rather than marginal (#11149).
+ "opencode-go",
]);
/**
diff --git a/src/lib/usage/internalUsageCommand.ts b/src/lib/usage/internalUsageCommand.ts
index 257b8f9c158..37f5f009d1c 100644
--- a/src/lib/usage/internalUsageCommand.ts
+++ b/src/lib/usage/internalUsageCommand.ts
@@ -13,7 +13,7 @@ const TEXT_PLAIN_HEADERS = { "Content-Type": "text/plain; charset=utf-8" } as co
type JsonRecord = Record;
-interface UsageCommandApiKeyMetadata {
+export interface UsageCommandApiKeyMetadata {
id: string;
name?: string;
allowedConnections?: string[] | null;
@@ -31,7 +31,7 @@ interface ProviderConnectionLike {
quotaWindowThresholds?: Record | null;
}
-interface UsageSnapshot {
+export interface UsageSnapshot {
connectionId: string;
provider: string;
plan: unknown;
@@ -39,7 +39,7 @@ interface UsageSnapshot {
quotaWindowThresholds?: Record | null;
}
-interface UsageCommandSelection {
+export interface UsageCommandSelection {
preferredProvider?: string | null;
preferredConnectionId?: string | null;
}
@@ -258,7 +258,7 @@ function snapshotFromConnection(
};
}
-async function collectUsageSnapshots(
+export async function collectUsageSnapshots(
metadata: UsageCommandApiKeyMetadata,
deps: RequiredDeps
): Promise {
@@ -525,6 +525,55 @@ function appendQuotaBlock(
lines.push(`⏱ reset in ${formatResetIn(getResetAt(match?.quota ?? null), now)}`);
}
+/**
+ * Structured form of the usage command — what {@link buildUsageCommandText}
+ * renders as text, exposed as data for API consumers (the OmniCopilot panel
+ * asks for it via `?format=json`). Text and JSON share the exact same
+ * collectors, so the two can never disagree about a number.
+ *
+ * The key design constraint is the 403 case: a key without `allowUsageCommand`
+ * must reach the client as a *structured* reason, not a bare text error — a
+ * caller rendering a usage panel has to be able to tell "the server does not
+ * know your limits yet" apart from "this key may not ask".
+ */
+/** Discriminated so the caller never reads a data field off a refusal:
+ * `allowed:false` carries only `error`; `allowed:true` carries the data. */
+export type UsageCommandJson =
+ | { allowed: false; error: { message: string } }
+ | {
+ allowed: true;
+ /** Present only when the key opted into per-key usage limits. */
+ personal: unknown | null;
+ /** The selected provider snapshot, or null when nothing is cached. */
+ provider: UsageSnapshot | null;
+ /** Every connection's snapshot, so a panel can render Codex / Claude /
+ * OpenCode side by side instead of only the selected one (#11191). The
+ * single-pick in `provider` is a presentation choice for a terminal; the
+ * collector already gathered all of them. */
+ providers: UsageSnapshot[];
+ };
+
+export async function buildUsageCommandJson(
+ metadata: UsageCommandApiKeyMetadata,
+ deps: InternalUsageCommandDeps = {},
+ selection: UsageCommandSelection = {}
+): Promise {
+ const resolvedDeps = await normalizeDeps(deps);
+ const personal =
+ metadata.usageLimitEnabled === true
+ ? await resolvedDeps.getApiKeyUsageLimitStatus(
+ {
+ ...metadata,
+ preferredProvider: selection.preferredProvider ?? metadata.preferredProvider ?? null,
+ },
+ { now: resolvedDeps.now }
+ )
+ : null;
+ const snapshots = await collectUsageSnapshots(metadata, resolvedDeps);
+ const provider = selectUsageSnapshot(snapshots, selection);
+ return { allowed: true, personal, provider, providers: snapshots };
+}
+
export async function buildUsageCommandText(
metadata: UsageCommandApiKeyMetadata,
deps: InternalUsageCommandDeps = {},
@@ -588,6 +637,17 @@ function inferHttpUsageCommandSelection(request: Request): UsageCommandSelection
}
}
+/** `?format=json` (or `?format=JSON`) — anything else falls back to the text
+ * form, which is the historical contract of this endpoint. */
+function wantsUsageCommandJson(request: Request): boolean {
+ try {
+ const format = new URL(request.url, "http://localhost").searchParams.get("format");
+ return format !== null && format.trim().toLowerCase() === "json";
+ } catch {
+ return false;
+ }
+}
+
function createPlainUsageCommandResponse(text: string, status = 200): Response {
return new Response(text, { status, headers: TEXT_PLAIN_HEADERS });
}
@@ -764,22 +824,45 @@ export async function handleInternalUsageCommandHttpRequest(
): Promise {
try {
const resolvedDeps = await normalizeDeps(deps);
+ const json = wantsUsageCommandJson(request);
const apiKey = extractUsageCommandApiKey(request);
if (!apiKey || !(await resolvedDeps.isValidApiKey(apiKey))) {
+ if (json) {
+ return Response.json(
+ { allowed: false, error: { message: USAGE_COMMAND_AUTH_REQUIRED_MESSAGE } } satisfies UsageCommandJson,
+ { status: 401 }
+ );
+ }
return createPlainUsageCommandResponse(USAGE_COMMAND_AUTH_REQUIRED_MESSAGE, 401);
}
const metadata = await resolvedDeps.getApiKeyMetadata(apiKey);
if (!metadata?.id) {
+ if (json) {
+ return Response.json(
+ { allowed: false, error: { message: USAGE_COMMAND_AUTH_REQUIRED_MESSAGE } } satisfies UsageCommandJson,
+ { status: 401 }
+ );
+ }
return createPlainUsageCommandResponse(USAGE_COMMAND_AUTH_REQUIRED_MESSAGE, 401);
}
if (metadata.allowUsageCommand !== true) {
+ if (json) {
+ return Response.json(
+ { allowed: false, error: { message: USAGE_COMMAND_DISABLED_MESSAGE } } satisfies UsageCommandJson,
+ { status: 403 }
+ );
+ }
return createPlainUsageCommandResponse(USAGE_COMMAND_DISABLED_MESSAGE, 403);
}
+ const selection = inferHttpUsageCommandSelection(request);
+ if (json) {
+ return Response.json(await buildUsageCommandJson(metadata, resolvedDeps, selection));
+ }
return createPlainUsageCommandResponse(
- await buildUsageCommandText(metadata, resolvedDeps, inferHttpUsageCommandSelection(request))
+ await buildUsageCommandText(metadata, resolvedDeps, selection)
);
} catch (err) {
const body = buildErrorBody(500, err instanceof Error ? err.message : String(err));
diff --git a/src/server/authz/pipeline.ts b/src/server/authz/pipeline.ts
index 9f4e46bed1e..d8dacf376d7 100644
--- a/src/server/authz/pipeline.ts
+++ b/src/server/authz/pipeline.ts
@@ -205,6 +205,7 @@ function drainingResponse(requestId: string): NextResponse {
{ status: 503 }
);
response.headers.set(AUTHZ_HEADER_REQUEST_ID, requestId);
+ response.headers.set("Retry-After", "5");
return response;
}
diff --git a/src/shared/components/ProviderIcon.tsx b/src/shared/components/ProviderIcon.tsx
index aeb6f777ba0..56d54bf0dc2 100644
--- a/src/shared/components/ProviderIcon.tsx
+++ b/src/shared/components/ProviderIcon.tsx
@@ -263,6 +263,7 @@ const KNOWN_PNGS = new Set([
"linkup-search",
"llamafile",
"llamagate",
+ "logfare",
"maritalk",
"nanobot",
"nanogpt",
diff --git a/src/shared/constants/providers.ts b/src/shared/constants/providers.ts
index 5c058958e76..b9b2b6951db 100644
--- a/src/shared/constants/providers.ts
+++ b/src/shared/constants/providers.ts
@@ -141,6 +141,7 @@ export const AGGREGATOR_PROVIDER_IDS = new Set([
"void-ai",
"helixmind",
"tabitoken",
+ "logfare",
]);
export const ENTERPRISE_CLOUD_PROVIDER_IDS = new Set([
diff --git a/src/shared/constants/providers/apikey/gateways.ts b/src/shared/constants/providers/apikey/gateways.ts
index 2baf13d3aae..6dab614425e 100644
--- a/src/shared/constants/providers/apikey/gateways.ts
+++ b/src/shared/constants/providers/apikey/gateways.ts
@@ -673,11 +673,12 @@ export const APIKEY_PROVIDERS_GATEWAYS = {
color: "#F97316",
textIcon: "G4F",
website: "https://g4f.space",
- hasFree: true,
- freeNote: "Free no-key reverse proxy to Groq (gpt4free project) — rate-limited to 5 req/min.",
+ hasFree: false,
+ freeNote:
+ "No-key reverse proxy to Groq (gpt4free project) — the anonymous free tier is gone; keyless calls return insufficient_credits until you bake proof-of-work credits. A g4f.dev member key is required.",
passthroughModels: true,
authHint:
- "No auth required. Free tier is limited to 5 requests/minute — sign up at g4f.dev/members.html for higher limits.",
+ "Anonymous use now needs proof-of-work credits baked at g4f.dev/chat — sign up at g4f.dev/members.html for a member key.",
},
"g4f-gemini": {
id: "g4f-gemini",
@@ -687,11 +688,12 @@ export const APIKEY_PROVIDERS_GATEWAYS = {
color: "#F97316",
textIcon: "G4F",
website: "https://g4f.space",
- hasFree: true,
- freeNote: "Free no-key reverse proxy to Gemini (gpt4free project) — rate-limited to 5 req/min.",
+ hasFree: false,
+ freeNote:
+ "No-key reverse proxy to Gemini (gpt4free project) — the anonymous free tier is gone; keyless calls return insufficient_credits until you bake proof-of-work credits. A g4f.dev member key is required.",
passthroughModels: true,
authHint:
- "No auth required. Free tier is limited to 5 requests/minute — sign up at g4f.dev/members.html for higher limits.",
+ "Anonymous use now needs proof-of-work credits baked at g4f.dev/chat — sign up at g4f.dev/members.html for a member key.",
},
"g4f-pollinations": {
id: "g4f-pollinations",
@@ -701,12 +703,12 @@ export const APIKEY_PROVIDERS_GATEWAYS = {
color: "#F97316",
textIcon: "G4F",
website: "https://g4f.space",
- hasFree: true,
+ hasFree: false,
freeNote:
- "Free no-key reverse proxy to Pollinations (gpt4free project) — rate-limited to 5 req/min.",
+ "No-key reverse proxy to Pollinations (gpt4free project) — the anonymous free tier is gone; keyless calls return insufficient_credits until you bake proof-of-work credits. A g4f.dev member key is required.",
passthroughModels: true,
authHint:
- "No auth required. Free tier is limited to 5 requests/minute — sign up at g4f.dev/members.html for higher limits.",
+ "Anonymous use now needs proof-of-work credits baked at g4f.dev/chat — sign up at g4f.dev/members.html for a member key.",
},
"g4f-ollama": {
id: "g4f-ollama",
@@ -716,11 +718,12 @@ export const APIKEY_PROVIDERS_GATEWAYS = {
color: "#F97316",
textIcon: "G4F",
website: "https://g4f.space",
- hasFree: true,
- freeNote: "Free no-key hosted Ollama gateway (gpt4free project) — rate-limited to 5 req/min.",
+ hasFree: false,
+ freeNote:
+ "No-key hosted Ollama gateway (gpt4free project) — the anonymous free tier is gone; keyless calls return insufficient_credits until you bake proof-of-work credits. A g4f.dev member key is required.",
passthroughModels: true,
authHint:
- "No auth required. Free tier is limited to 5 requests/minute — sign up at g4f.dev/members.html for higher limits.",
+ "Anonymous use now needs proof-of-work credits baked at g4f.dev/chat — sign up at g4f.dev/members.html for a member key.",
},
"g4f-nvidia": {
id: "g4f-nvidia",
@@ -730,12 +733,12 @@ export const APIKEY_PROVIDERS_GATEWAYS = {
color: "#F97316",
textIcon: "G4F",
website: "https://g4f.space",
- hasFree: true,
+ hasFree: false,
freeNote:
- "Free no-key reverse proxy to NVIDIA NIM (gpt4free project) — rate-limited to 5 req/min.",
+ "No-key reverse proxy to NVIDIA NIM (gpt4free project) — the anonymous free tier is gone; keyless calls return insufficient_credits until you bake proof-of-work credits. A g4f.dev member key is required.",
passthroughModels: true,
authHint:
- "No auth required. Free tier is limited to 5 requests/minute — sign up at g4f.dev/members.html for higher limits.",
+ "Anonymous use now needs proof-of-work credits baked at g4f.dev/chat — sign up at g4f.dev/members.html for a member key.",
},
"vercel-ai-gateway": {
id: "vercel-ai-gateway",
@@ -1265,6 +1268,29 @@ export const APIKEY_PROVIDERS_GATEWAYS = {
apiHint:
"Create a helix- key and use https://helixmind.online/v1. OpenAI requests use Bearer authentication; the Anthropic-compatible messages endpoint accepts x-api-key.",
},
+ // Logfare (https://logfare.ai) — free OpenAI-compatible inference, live-verified
+ // 2026-08-21 (real /v1/models catalog; 11 chat-capable models incl. kimi-k3,
+ // deepseek-v4-pro, glm-5.2, gpt-5.6-luna). Key issued instantly at /register
+ // (username/password, no email). ⚠️ Logfare logs every request in exchange for
+ // free inference (opt out at /consent) — surfaced in freeNote per the catalog
+ // convention for data-collecting free providers.
+ logfare: {
+ id: "logfare",
+ alias: "logfare",
+ name: "Logfare",
+ icon: "auto_awesome",
+ color: "#22C55E",
+ textIcon: "LF",
+ website: "https://logfare.ai",
+ hasFree: true,
+ freeNote:
+ "Free OpenAI-compatible inference — no rate limits, no card. Logfare logs every request (prompts, completions, metadata) for internal research; opt out at /consent. Read https://logfare.ai/tos and https://logfare.ai/privacy before use.",
+ authHint:
+ "Create a free account at https://logfare.ai/register (username/password, no email verification) to get an instant API key, then paste it here as a Bearer token.",
+ apiHint:
+ "Create a free API key at https://logfare.ai/register, then use https://logfare.ai/v1 as the OpenAI-compatible base URL. Note the request-logging policy: prompts, completions and metadata are logged for research (opt out at https://logfare.ai/consent).",
+ passthroughModels: true,
+ },
// TabiToken (https://tabitoken.com) — NewAPI-based Claude gateway. Its public pricing
// endpoint lists a Claude-only catalog (Opus 5 / 4.8, each with a -thinking variant),
// every model accepting the Anthropic and OpenAI protocols.
diff --git a/src/shared/middleware/chatBodyAdmission.ts b/src/shared/middleware/chatBodyAdmission.ts
index 1c3c25b904b..9b20d78e5ae 100644
--- a/src/shared/middleware/chatBodyAdmission.ts
+++ b/src/shared/middleware/chatBodyAdmission.ts
@@ -18,6 +18,7 @@
import { CORS_HEADERS } from "../utils/cors";
import { createHmac } from "crypto";
import v8 from "node:v8";
+import { trackRequest } from "../../lib/gracefulShutdown";
function parsePositiveInt(value: string | undefined, fallback: number): number {
const parsed = Number.parseInt(String(value), 10);
@@ -229,6 +230,7 @@ export class ChatAdmissionController {
tryAcquireHealthyHeadroom(): ChatAdmissionLease | null {
if (this.#activeHealthy >= this.healthyHeadroom) return null;
this.#activeHealthy += 1;
+ const done = trackRequest();
let released = false;
return {
get released() {
@@ -238,6 +240,7 @@ export class ChatAdmissionController {
if (released) return;
released = true;
this.#activeHealthy = Math.max(0, this.#activeHealthy - 1);
+ done();
},
};
}
@@ -264,6 +267,7 @@ export class ChatAdmissionController {
tryAcquireHeavy(): ChatAdmissionLease | null {
if (this.#activeHeavy >= this.maxHeavyInFlight) return null;
this.#activeHeavy += 1;
+ const done = trackRequest();
let released = false;
return {
get released() {
@@ -273,6 +277,7 @@ export class ChatAdmissionController {
if (released) return;
released = true;
this.#activeHeavy = Math.max(0, this.#activeHeavy - 1);
+ done();
this.#dispatchFair();
},
};
diff --git a/src/shared/validation/schemas/combo.ts b/src/shared/validation/schemas/combo.ts
index edeca47e729..db825e11cc9 100644
--- a/src/shared/validation/schemas/combo.ts
+++ b/src/shared/validation/schemas/combo.ts
@@ -321,7 +321,7 @@ export const createComboSchema = z
.object({
name: comboNameSchema,
description: z.string().max(2000).optional(),
- models: z.array(comboModelEntry).optional().default([]),
+ models: z.array(comboModelEntry).min(1, "a combo requires at least one model"),
strategy: comboStrategySchema.optional().default("priority"),
config: comboRuntimeConfigSchema.optional(),
allowedProviders: z.array(z.string().trim().min(1).max(200)).max(100).optional(),
@@ -380,8 +380,9 @@ export const updateComboSchema = z
.object({
name: comboNameSchema.optional(),
description: z.string().max(2000).optional().nullable(),
- // Creation may leave `models` empty (`omniroute combo create` drafts one
- // that way); an update may not, or a working combo loses every target.
+ // An update may not remove every model from a combo, or a working combo
+ // loses every target. Creation refuses an empty list too: since the CLI
+ // gained --models (#10954), an empty draft has no remaining legitimate path.
models: z
.array(comboModelEntry)
.min(1, "an update cannot remove every model from a combo")
diff --git a/src/sse/handlers/chat.ts b/src/sse/handlers/chat.ts
index 3ec6e6c6d6c..aadf7d29d3d 100644
--- a/src/sse/handlers/chat.ts
+++ b/src/sse/handlers/chat.ts
@@ -1851,7 +1851,7 @@ async function handleSingleModelChat(
modelPinned: runtimeOptions?.modelPinned ?? false,
routingComboId: runtimeOptions?.routingComboId ?? null,
sessionAffinityKey: runtimeOptions.sessionAffinityKey ?? null,
- reasoningTransportFallback: runtimeOptions.reasoningTransportFallback ?? "skip",
+ reasoningTransportFallback: runtimeOptions.reasoningTransportFallback ?? "drop",
managedLease: runtimeOptions.managedLease ?? null,
},
runtimeOptions
@@ -2240,6 +2240,7 @@ async function handleSingleModelChat(
if (
!runtimeOptions.emergencyFallbackTried &&
!comboName &&
+ !forceLiveComboTest &&
shouldRetrySameAccountTransport({
status: result.status,
errorText: errorStr,
diff --git a/src/sse/handlers/chatHelpers.ts b/src/sse/handlers/chatHelpers.ts
index 36bd4077408..bcf9a51c62c 100644
--- a/src/sse/handlers/chatHelpers.ts
+++ b/src/sse/handlers/chatHelpers.ts
@@ -422,7 +422,7 @@ export async function executeChatWithBreaker({
conversationId = null,
modelPinned = false,
routingComboId = null,
- reasoningTransportFallback = "skip",
+ reasoningTransportFallback = "drop",
sessionAffinityKey = null,
managedLease = null,
}: ExecuteChatWithBreakerOptions): Promise {
diff --git a/src/sse/services/auth.ts b/src/sse/services/auth.ts
index b1df29552a4..5d3d8851786 100644
--- a/src/sse/services/auth.ts
+++ b/src/sse/services/auth.ts
@@ -106,8 +106,17 @@ import {
resolveProviderId,
NOAUTH_PROVIDERS,
WEB_COOKIE_PROVIDERS,
+ isSelfHostedChatProvider,
} from "@/shared/constants/providers";
-import { isModelExcludedByConnection } from "@/domain/connectionModelRules";
+import {
+ isModelExcludedByConnection,
+ isModelAdvertisedByConnection,
+} from "@/domain/connectionModelRules";
+import {
+ getSyncedAvailableModelsByConnection,
+ SYNCED_AVAILABLE_MODELS_MALFORMED,
+ type SyncedAvailableModelsByConnection,
+} from "@/lib/db/models";
import { isFreeModel } from "@/shared/utils/freeModels";
import {
applySessionAffinityPin,
@@ -1160,6 +1169,54 @@ function materializeConnection(
};
}
+/**
+ * #11089: load the per-connection synced model inventory for self-hosted chat
+ * providers so connection selection can drop hosts that never advertised the
+ * requested model.
+ *
+ * Scoped to SELF_HOSTED_CHAT_PROVIDER_IDS: those are the providers where one
+ * provider id fans out to several independent hosts with genuinely different
+ * inventories. Hosted providers share one catalog per provider, so filtering
+ * there would only add a DB read.
+ *
+ * Returns an empty map (= no filtering) when there is no model to match, when
+ * no candidate is self-hosted, or when the persisted rows are malformed — a
+ * partial read must never silently shrink the pool.
+ */
+async function loadAdvertisedModelsForSelfHostedConnections(
+ connections: ProviderConnectionView[],
+ requestedModel: string | null
+): Promise>> {
+ const advertised = new Map>();
+ if (!requestedModel) return advertised;
+
+ const selfHostedProviders = new Set(
+ connections
+ .map((c) => c.provider)
+ .filter((p): p is string => typeof p === "string" && isSelfHostedChatProvider(p))
+ );
+ if (selfHostedProviders.size === 0) return advertised;
+
+ await Promise.all(
+ [...selfHostedProviders].map(async (providerId) => {
+ let byConnection: SyncedAvailableModelsByConnection;
+ try {
+ byConnection = await getSyncedAvailableModelsByConnection(providerId);
+ } catch {
+ return;
+ }
+ // Malformed persisted rows: fail open for the whole provider.
+ if (byConnection[SYNCED_AVAILABLE_MODELS_MALFORMED]) return;
+ for (const [connectionId, models] of Object.entries(byConnection)) {
+ if (!Array.isArray(models) || models.length === 0) continue;
+ advertised.set(connectionId, new Set(models.map((m) => m.id)));
+ }
+ })
+ );
+
+ return advertised;
+}
+
/**
* Get provider credentials from localDb
* Filters out unavailable accounts and returns the selected account based on strategy
@@ -1435,6 +1492,14 @@ export async function getProviderCredentials(
let modelLockedCount = 0;
let familyLockedCount = 0;
const connectionFilterStatus = new Map();
+ // #11089: multi-host self-hosted providers keep a per-connection synced
+ // inventory. Without it, a request can be routed to a host that never had
+ // the model, producing a spurious model-not-found instead of pinning to
+ // the host that does. Empty map = no inventory known = no filtering.
+ const advertisedModelsByConnection = await loadAdvertisedModelsForSelfHostedConnections(
+ connections,
+ requestedModel
+ );
// Filter out unavailable accounts and excluded connection
let availableConnections = connections.filter((c) => {
if (excludedConnectionIds.has(c.id)) {
@@ -1445,6 +1510,13 @@ export async function getProviderCredentials(
connectionFilterStatus.set(c.id, "modelExcluded");
return false;
}
+ if (
+ requestedModel &&
+ !isModelAdvertisedByConnection(requestedModel, advertisedModelsByConnection.get(c.id))
+ ) {
+ connectionFilterStatus.set(c.id, "modelNotAdvertised");
+ return false;
+ }
if (!allowSuppressedConnections) {
if (!allowRateLimitedConnections && isAccountUnavailable(c.rateLimitedUntil)) {
connectionFilterStatus.set(c.id, "rateLimited");
@@ -1510,6 +1582,7 @@ export async function getProviderCredentials(
const codexScopeLimited = status === "codexScopeLimited";
const modelLocked = status === "modelLocked";
const modelExcluded = status === "modelExcluded";
+ const modelNotAdvertised = status === "modelNotAdvertised";
if (excluded || rateLimited) {
log.debug(
"AUTH",
@@ -1520,6 +1593,11 @@ export async function getProviderCredentials(
"AUTH",
` → ${c.id?.slice(0, 8)} | excluded by per-account model rule for ${requestedModel}`
);
+ } else if (modelNotAdvertised) {
+ log.debug(
+ "AUTH",
+ ` → ${c.id?.slice(0, 8)} | synced inventory does not advertise ${requestedModel}`
+ );
} else if (terminalStatus) {
log.debug(
"AUTH",
@@ -3104,11 +3182,9 @@ export async function clearAccountError(
}
/**
- * Optional CAS token. When provided, the clear is performed via an atomic
- * conditional UPDATE (clearConnectionErrorIfUnchanged) that aborts if the row
- * was written by a concurrent path between the caller's snapshot read and this
- * clear. Closes the TOCTOU window in the quota-recovery path. When omitted,
- * the clear is unconditional (preserves existing post-success-call behavior).
+ * Optional CAS token. When provided, clearConnectionErrorIfUnchanged atomically
+ * aborts if another path modified the row after the caller's snapshot.
+ * This closes the TOCTOU window; omission preserves unconditional clearing.
*/
export interface RecoveredStateExpectation {
testStatus: string | null;
@@ -3116,23 +3192,25 @@ export interface RecoveredStateExpectation {
rateLimitedUntil: string | null;
}
export async function clearRecoveredProviderState(
- credentials: Partial | null,
+ credentials: unknown,
expectedState?: RecoveredStateExpectation
): Promise<{ applied: boolean }> {
- if (!credentials?.connectionId) return { applied: false };
+ const recoverable = credentials as Partial | null;
+ if (typeof recoverable?.connectionId !== "string" || !recoverable.connectionId)
+ return { applied: false };
if (expectedState) {
- const applied = await clearConnectionErrorIfUnchanged(credentials.connectionId, expectedState);
+ const applied = await clearConnectionErrorIfUnchanged(recoverable.connectionId, expectedState);
if (!applied) {
log.info(
"AUTH",
- `Skipped recovery clear for ${credentials.connectionId.slice(0, 8)} — state changed concurrently (CAS miss)`
+ `Skipped recovery clear for ${recoverable.connectionId.slice(0, 8)} — state changed concurrently (CAS miss)`
);
return { applied: false };
}
- log.info("AUTH", `Account ${credentials.connectionId.slice(0, 8)} error cleared (CAS)`);
+ log.info("AUTH", `Account ${recoverable.connectionId.slice(0, 8)} error cleared (CAS)`);
return { applied: true };
}
- await clearAccountError(credentials.connectionId, credentials);
+ await clearAccountError(recoverable.connectionId, recoverable);
return { applied: true };
}
type AuthRequestLike = {
diff --git a/tests/snapshots/provider/translate-path.json b/tests/snapshots/provider/translate-path.json
index e371bbc8581..7a6b93390a0 100644
--- a/tests/snapshots/provider/translate-path.json
+++ b/tests/snapshots/provider/translate-path.json
@@ -3585,6 +3585,29 @@
"stream": "https://arena.ai/nextjs-api/stream/create-evaluation"
}
},
+ "logfare": {
+ "format": "openai",
+ "headers": {
+ "apiKey": {
+ "Accept": "text/event-stream",
+ "Authorization": "Bearer ",
+ "Content-Type": "application/json"
+ },
+ "nonStream": {
+ "Authorization": "Bearer ",
+ "Content-Type": "application/json"
+ },
+ "oauth": {
+ "Accept": "text/event-stream",
+ "Authorization": "Bearer ",
+ "Content-Type": "application/json"
+ }
+ },
+ "url": {
+ "nonStream": "https://logfare.ai/v1/chat/completions",
+ "stream": "https://logfare.ai/v1/chat/completions"
+ }
+ },
"longcat": {
"format": "openai",
"headers": {
diff --git a/tests/unit/account-rotation-lot-c.test.ts b/tests/unit/account-rotation-lot-c.test.ts
index 0dde306f90a..6ae41c42ed8 100644
--- a/tests/unit/account-rotation-lot-c.test.ts
+++ b/tests/unit/account-rotation-lot-c.test.ts
@@ -1,6 +1,11 @@
import { test } from "node:test";
import assert from "node:assert/strict";
-import { pickAccount, markCooldown, markSuccess, isAccountReady } from "../../open-sse/executors/accountRotation.ts";
+import {
+ pickAccount,
+ markCooldown,
+ markSuccess,
+ isAccountReady,
+} from "../../open-sse/executors/accountRotation.ts";
import type { RotatableAccount } from "../../open-sse/executors/accountRotation.ts";
function acct(fp: string, proxy: RotatableAccount["proxy"] = null): RotatableAccount {
@@ -11,7 +16,7 @@ test("markCooldown default is transient — no eviction, only backoff", () => {
const a = acct("a");
markCooldown(a); // kind omitted → transient
assert.ok(a.cooldownUntil > Date.now());
- assert.equal((a as Record).evictedAt, undefined);
+ assert.equal(a.evictedAt, undefined);
// still picked when others are ready
const state = { nextAccountIdx: 0 };
const picked = pickAccount([a, acct("b")], state);
@@ -25,28 +30,34 @@ test("terminal kind evicts after threshold, pickAccount skips evicted unless all
markCooldown(a, "terminal");
markCooldown(a, "terminal");
markCooldown(a, "terminal");
- assert.ok((a as Record).evictedAt != null);
+ assert.ok(a.evictedAt != null);
const state = { nextAccountIdx: 0 };
// b is ready, a evicted → b is picked
- const picked = pickAccount([a, b], state, (x) => isAccountReady(x) && !(x as Record).evictedAt);
+ const picked = pickAccount([a, b], state, (x) => isAccountReady(x) && !x.evictedAt);
assert.equal(picked.fingerprint, "healthy");
// when all evicted, caller still gets an account rather than hanging (preserves :52-58)
- (b as Record).evictedAt = Date.now();
- const fallback = pickAccount([a, b], { nextAccountIdx: 0 }, (x) => isAccountReady(x) && !(x as Record).evictedAt);
+ b.evictedAt = Date.now();
+ const fallback = pickAccount(
+ [a, b],
+ { nextAccountIdx: 0 },
+ (x) => isAccountReady(x) && !x.evictedAt
+ );
assert.ok(fallback.fingerprint === "dead" || fallback.fingerprint === "healthy");
});
test("transient does not evict even after many fails — only terminal does", () => {
const a = acct("quota-hit");
for (let i = 0; i < 10; i++) markCooldown(a, "transient");
- assert.equal((a as Record).evictedAt, undefined);
+ assert.equal(a.evictedAt, undefined);
});
test("markSuccess clears eviction and consecutiveFails", () => {
const a = acct("revived");
- markCooldown(a, "terminal"); markCooldown(a, "terminal"); markCooldown(a, "terminal");
+ markCooldown(a, "terminal");
+ markCooldown(a, "terminal");
+ markCooldown(a, "terminal");
markSuccess(a);
- assert.equal((a as Record).evictedAt, null);
+ assert.equal(a.evictedAt, null);
assert.equal(a.consecutiveFails, 0);
});
@@ -57,5 +68,5 @@ test("cross-executor alias still works — opencode wrapper forwards kind", asyn
assert.ok(mc.length >= 1 && mc.length <= 2);
// Prove it accepts terminal without throw
const tmp = acct("probe");
- assert.doesNotThrow(() => (mc as Record)(tmp, "terminal"));
+ assert.doesNotThrow(() => mc(tmp, "terminal"));
});
diff --git a/tests/unit/account-rotation.test.ts b/tests/unit/account-rotation.test.ts
index f4864ee2dd8..a682f02a0d8 100644
--- a/tests/unit/account-rotation.test.ts
+++ b/tests/unit/account-rotation.test.ts
@@ -7,6 +7,8 @@ import {
markSuccess,
maskAccountId,
isNetworkErrorRotatable,
+ isEmptyUpstreamRejection,
+ extractChatcmplId,
type RotatableAccount,
} from "../../open-sse/executors/accountRotation.ts";
@@ -114,3 +116,100 @@ describe("accountRotation", () => {
assert.strictEqual(isNetworkErrorRotatable(withoutProxy), false);
});
});
+
+describe("isEmptyUpstreamRejection", () => {
+ it("matches the observed malformed completion envelope (no error field, empty content, null finish_reason)", () => {
+ const observed =
+ '{"id":"chatcmpl_44fn2g6e7kk","object":"chat.completion","created":1787419957,"model":"muse-spark-1.2-contributor-free","choices":[{"index":0,"message":{"role":"assistant"},"finish_reason":null}]}';
+ assert.strictEqual(isEmptyUpstreamRejection(400, observed), true);
+ });
+
+ it("does not match a non-400 status", () => {
+ const observed =
+ '{"id":"chatcmpl_44fn2g6e7kk","object":"chat.completion","created":1787419957,"model":"muse-spark-1.2-contributor-free","choices":[{"index":0,"message":{"role":"assistant"},"finish_reason":null}]}';
+ assert.strictEqual(isEmptyUpstreamRejection(200, observed), false);
+ assert.strictEqual(isEmptyUpstreamRejection(429, observed), false);
+ assert.strictEqual(isEmptyUpstreamRejection(502, observed), false);
+ });
+
+ it("does not match when an error field is present", () => {
+ const withError = JSON.stringify({
+ error: { message: "bad request", type: "invalid_request_error" },
+ });
+ assert.strictEqual(isEmptyUpstreamRejection(400, withError), false);
+ const emptyError = JSON.stringify({ error: {} });
+ assert.strictEqual(isEmptyUpstreamRejection(400, emptyError), false);
+ });
+
+ it("does not match when content is non-empty or tool_calls present", () => {
+ const nonEmpty = JSON.stringify({
+ choices: [{ message: { role: "assistant", content: "hi" }, finish_reason: "stop" }],
+ });
+ assert.strictEqual(isEmptyUpstreamRejection(400, nonEmpty), false);
+
+ const toolCalls = JSON.stringify({
+ choices: [
+ { message: { role: "assistant", tool_calls: [{ id: "x" }] }, finish_reason: "tool_calls" },
+ ],
+ });
+ assert.strictEqual(isEmptyUpstreamRejection(400, toolCalls), false);
+ });
+
+ it("does not match when content is a non-string non-null value (number, block array)", () => {
+ const numericContent = JSON.stringify({
+ choices: [{ message: { role: "assistant", content: 123 }, finish_reason: null }],
+ });
+ assert.strictEqual(
+ isEmptyUpstreamRejection(400, numericContent),
+ false,
+ "non-string non-null content is not eligible"
+ );
+
+ const reasoningContent = JSON.stringify({
+ choices: [
+ { message: { role: "assistant", reasoning_content: "thinking" }, finish_reason: null },
+ ],
+ });
+ assert.strictEqual(isEmptyUpstreamRejection(400, reasoningContent), false);
+ });
+
+ it("does not match when choices or message are absent", () => {
+ const noChoices = JSON.stringify({ id: "chatcmpl_x", model: "muse" });
+ assert.strictEqual(isEmptyUpstreamRejection(400, noChoices), false);
+ const noMessage = JSON.stringify({ choices: [{ finish_reason: null }] });
+ assert.strictEqual(isEmptyUpstreamRejection(400, noMessage), false);
+ });
+
+ it("does not match when finish_reason is a literal value (not null)", () => {
+ const stopReason = JSON.stringify({
+ choices: [{ message: { role: "assistant" }, finish_reason: "stop" }],
+ });
+ assert.strictEqual(isEmptyUpstreamRejection(400, stopReason), false);
+ });
+
+ it("matches an empty string content (treated as eligible)", () => {
+ const emptyContent = JSON.stringify({
+ choices: [{ message: { role: "assistant", content: "" }, finish_reason: null }],
+ });
+ assert.strictEqual(isEmptyUpstreamRejection(400, emptyContent), true);
+ });
+
+ it("returns false for unparseable JSON rather than throwing", () => {
+ assert.strictEqual(isEmptyUpstreamRejection(400, "not json"), false);
+ assert.strictEqual(isEmptyUpstreamRejection(400, ""), false);
+ });
+});
+
+describe("extractChatcmplId", () => {
+ it("extracts the chatcmpl id from an observed envelope", () => {
+ const observed =
+ '{"id":"chatcmpl_44fn2g6e7kk","object":"chat.completion","created":1787419957,"model":"muse-spark-1.2-contributor-free","choices":[{"index":0,"message":{"role":"assistant"},"finish_reason":null}]}';
+ assert.strictEqual(extractChatcmplId(observed), "chatcmpl_44fn2g6e7kk");
+ });
+
+ it("falls back to 'unknown' when no id is present", () => {
+ assert.strictEqual(extractChatcmplId("{choices:[]}"), "unknown");
+ assert.strictEqual(extractChatcmplId(""), "unknown");
+ assert.strictEqual(extractChatcmplId("not json"), "unknown");
+ });
+});
diff --git a/tests/unit/antigravity-dynamic-session-id-10443.test.ts b/tests/unit/antigravity-dynamic-session-id-10443.test.ts
new file mode 100644
index 00000000000..af5d198ff9b
--- /dev/null
+++ b/tests/unit/antigravity-dynamic-session-id-10443.test.ts
@@ -0,0 +1,18 @@
+import assert from "node:assert/strict";
+import { test } from "node:test";
+import { getAntigravitySessionId } from "../../open-sse/services/antigravityIdentity.ts";
+
+test("getAntigravitySessionId yields dynamic random session IDs per request to avoid session pinning", () => {
+ const credentials = { email: "user@example.com", connectionId: "conn_123" };
+
+ const id1 = getAntigravitySessionId(credentials);
+ const id2 = getAntigravitySessionId(credentials);
+
+ assert.notEqual(id1, id2, "getAntigravitySessionId should not pin to a static account email hash");
+ assert.equal(typeof id1, "string");
+ assert.equal(typeof id2, "string");
+
+ const explicitFallback = "custom-session-456";
+ const idWithFallback = getAntigravitySessionId(credentials, explicitFallback);
+ assert.equal(idWithFallback, explicitFallback, "explicit fallback session ID should take precedence");
+});
diff --git a/tests/unit/auth-clear-account-error.test.ts b/tests/unit/auth-clear-account-error.test.ts
index 1131ae63701..a303da4e7ed 100644
--- a/tests/unit/auth-clear-account-error.test.ts
+++ b/tests/unit/auth-clear-account-error.test.ts
@@ -104,6 +104,11 @@ test("clearRecoveredProviderState ignores empty payloads and clears recoverable
await auth.clearRecoveredProviderState(null);
await auth.clearRecoveredProviderState({});
+ await auth.clearRecoveredProviderState({
+ allExpired: true,
+ expiredCount: 1,
+ expiredStatus: "expired",
+ });
await auth.clearRecoveredProviderState({
connectionId: created.id,
testStatus: "unavailable",
diff --git a/tests/unit/authz/pipeline.test.ts b/tests/unit/authz/pipeline.test.ts
index 34703f270ec..6f6469e804e 100644
--- a/tests/unit/authz/pipeline.test.ts
+++ b/tests/unit/authz/pipeline.test.ts
@@ -306,6 +306,7 @@ test("runAuthzPipeline rejects new API requests during shutdown drain", async ()
assert.equal(response.status, 503);
assert.equal(body.error.code, "SERVICE_UNAVAILABLE");
+ assert.equal(response.headers.get("retry-after"), "5");
});
test("runAuthzPipeline rejects rewritten API aliases during shutdown drain", async () => {
@@ -319,6 +320,7 @@ test("runAuthzPipeline rejects rewritten API aliases during shutdown drain", asy
assert.equal(response.status, 503);
assert.equal(response.headers.get("x-omniroute-route-class"), "CLIENT_API");
assert.equal(body.error.code, "SERVICE_UNAVAILABLE");
+ assert.equal(response.headers.get("retry-after"), "5");
});
test("runAuthzPipeline allows dashboard sessions to read model catalog aliases", async () => {
diff --git a/tests/unit/auto-keyless-custom-provider-11180.test.ts b/tests/unit/auto-keyless-custom-provider-11180.test.ts
new file mode 100644
index 00000000000..c2e97e8158d
--- /dev/null
+++ b/tests/unit/auto-keyless-custom-provider-11180.test.ts
@@ -0,0 +1,91 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+import fs from "node:fs";
+import os from "node:os";
+import path from "node:path";
+
+// #11180 regression guard: a custom OpenAI-compatible connection pointing at a
+// keyless local backend (llama.cpp / Ollama / vLLM started without an API key)
+// carries no apiKey, no OAuth token and no provider-specific session data, so
+// `hasUsableConnectionCredential` dropped it from `validConnections` before the
+// auto/* candidate pool was built. The connection was active, tested and synced,
+// yet structurally invisible to auto-routing with no log line and no UI hint.
+//
+// Keyless is the NORMAL configuration for a self-hosted backend, so a custom
+// compatible connection must stay eligible. This gate is one step later than
+// #5873 (registry-absent defaultModel fallback), whose guard still passes.
+
+const TEST_DATA_DIR = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-auto-keyless-11180-"));
+const ORIGINAL_DATA_DIR = process.env.DATA_DIR;
+
+process.env.DATA_DIR = TEST_DATA_DIR;
+
+const core = await import("../../src/lib/db/core.ts");
+const providersDb = await import("../../src/lib/db/providers.ts");
+const virtualFactory = await import("../../open-sse/services/autoCombo/virtualFactory.ts");
+
+type VirtualComboResult = Awaited>;
+
+async function resetStorage() {
+ core.resetDbInstance();
+ fs.rmSync(TEST_DATA_DIR, { recursive: true, force: true });
+ fs.mkdirSync(TEST_DATA_DIR, { recursive: true });
+}
+
+test.beforeEach(async () => {
+ await resetStorage();
+});
+
+test.after(async () => {
+ await resetStorage();
+ fs.rmSync(TEST_DATA_DIR, { recursive: true, force: true });
+
+ if (ORIGINAL_DATA_DIR === undefined) {
+ delete process.env.DATA_DIR;
+ } else {
+ process.env.DATA_DIR = ORIGINAL_DATA_DIR;
+ }
+});
+
+test("keyless custom openai-compatible connection enters the auto pool (#11180)", async () => {
+ const customProvider = "openai-compatible-chat-c2fe8a44-f2fd-47b4-8893-6f1521804c45";
+ await providersDb.createProviderConnection({
+ provider: customProvider,
+ authType: "apikey",
+ name: "llamaAsimov",
+ // Keyless local backend: llama-server --host 0.0.0.0 with no --api-key.
+ apiKey: "",
+ defaultModel: "Qwen3.8-27B-UD-Q4-DFlash-GGUF",
+ });
+
+ const combo: VirtualComboResult = await virtualFactory.createVirtualAutoCombo("fast");
+
+ const candidate = combo.models.find((model) => model.providerId === customProvider);
+ assert.ok(
+ candidate,
+ "a keyless custom-compatible connection must not be dropped by the credential gate"
+ );
+ assert.equal(candidate.model, `${customProvider}/Qwen3.8-27B-UD-Q4-DFlash-GGUF`);
+ assert.ok(combo.autoConfig.candidatePool.includes(customProvider));
+});
+
+test("a keyless FIRST-PARTY provider connection stays out of the pool (#11180)", async () => {
+ // The relaxation is scoped to custom compatible connection IDs. A first-party
+ // provider with an empty key is an unconfigured connection, not a keyless
+ // local backend, and must still be filtered out.
+ await providersDb.createProviderConnection({
+ provider: "openai",
+ authType: "apikey",
+ name: "unconfigured openai",
+ apiKey: "",
+ defaultModel: "gpt-4o",
+ });
+
+ const combo: VirtualComboResult = await virtualFactory.createVirtualAutoCombo("fast");
+
+ assert.equal(
+ combo.autoConfig.candidatePool.includes("openai"),
+ false,
+ "an unconfigured first-party connection must remain excluded"
+ );
+});
diff --git a/tests/unit/capture-critical-db-state.test.ts b/tests/unit/capture-critical-db-state.test.ts
index fcf2a017b7e..957c1afce36 100644
--- a/tests/unit/capture-critical-db-state.test.ts
+++ b/tests/unit/capture-critical-db-state.test.ts
@@ -4,6 +4,8 @@ import fs from "node:fs";
import os from "node:os";
import path from "node:path";
+type CoreModule = typeof import("../../src/lib/db/core.ts");
+
// Shared across all tests — the module caches DATA_DIR / SQLITE_FILE at load time,
// so we must create the temp dir and import exactly once.
type CoreModule = typeof import("../../src/lib/db/core.ts");
diff --git a/tests/unit/cc-compatible-provider.test.ts b/tests/unit/cc-compatible-provider.test.ts
index 9784382ca38..cbb16a28194 100644
--- a/tests/unit/cc-compatible-provider.test.ts
+++ b/tests/unit/cc-compatible-provider.test.ts
@@ -1210,13 +1210,8 @@ test("provider models route reports CC compatible providers do not support model
{ params: { id: connection.id } }
);
- assert.ok(
- response.status === 400 || response.status === 200,
- `CC-compatible models route should 400 (unsupported) or 200 (listed), got ${response.status}`
- );
- if (response.status === 400) {
- assert.deepEqual(await response.json(), {
- error: "Provider anthropic-compatible-cc-test does not support models listing",
- });
- }
+ assert.equal(response.status, 400);
+ assert.deepEqual(await response.json(), {
+ error: "Provider anthropic-compatible-cc-test does not support models listing",
+ });
});
diff --git a/tests/unit/chat-body-admission.test.ts b/tests/unit/chat-body-admission.test.ts
index e4707402a15..05afae869c8 100644
--- a/tests/unit/chat-body-admission.test.ts
+++ b/tests/unit/chat-body-admission.test.ts
@@ -16,6 +16,7 @@ const {
resolveSelfLoopBearer,
} = admissionModule;
const { withEarlyStreamKeepalive } = await import("../../open-sse/utils/earlyStreamKeepalive.ts");
+const { getActiveRequestCount } = await import("../../src/lib/gracefulShutdown.ts");
/**
* Save/restore the env-var keys that `resolveSelfLoopBearer` reads so tests can
@@ -49,6 +50,25 @@ function chatRequest(body: string, contentLength: string | null = String(body.le
});
}
+test("heavyweight leases are counted for SIGTERM drain (#11015)", () => {
+ globalThis.__omnirouteShutdown = { init: true, shuttingDown: false, activeRequests: 0 };
+ const controller = new ChatAdmissionController(2);
+ const before = getActiveRequestCount();
+ const lease = controller.tryAcquireHeavy();
+ assert.ok(lease);
+ assert.equal(getActiveRequestCount(), before + 1);
+ const headroom = controller.tryAcquireHealthyHeadroom();
+ assert.ok(headroom);
+ assert.equal(getActiveRequestCount(), before + 2);
+ lease.release();
+ assert.equal(getActiveRequestCount(), before + 1);
+ headroom.release();
+ assert.equal(getActiveRequestCount(), before);
+ lease.release();
+ headroom.release();
+ assert.equal(getActiveRequestCount(), before);
+});
+
test("small known body is admitted without consuming heavyweight capacity", async () => {
const controller = new ChatAdmissionController(1);
const result = await admitChatRequest(chatRequest("{}"), {
diff --git a/tests/unit/chat-routing-synced-inventory-11089.test.ts b/tests/unit/chat-routing-synced-inventory-11089.test.ts
new file mode 100644
index 00000000000..441b14893b6
--- /dev/null
+++ b/tests/unit/chat-routing-synced-inventory-11089.test.ts
@@ -0,0 +1,184 @@
+/**
+ * tests/unit/chat-routing-synced-inventory-11089.test.ts
+ *
+ * #11089 — Chat routing ignores per-connection model inventory on multi-host
+ * local providers.
+ *
+ * One self-hosted provider (`ollama-local`) with TWO connections pointing at
+ * different hosts and DISJOINT synced inventories:
+ *
+ * studio (priority 1) → gemma3:4b, flux2-klein:9b
+ * jetson (priority 2) → gemma3:4b
+ *
+ * `getProviderCredentials` only ever consulted the manual `excludedModels`
+ * denylist, never the synced inventory persisted per connection, so a request
+ * for `flux2-klein:9b` could land on jetson — a host that never had the model.
+ *
+ * Cases:
+ * 1. Model advertised by only one connection → the other is never selected.
+ * 2. Higher-priority host cooling → must NOT preemptively fall to a host that
+ * lacks the model.
+ * 3. The advertising connection stays selectable.
+ * 4. Model advertised by both → both remain eligible (no over-filtering).
+ * 5. Provider with NO synced inventory at all → fail open, selection unchanged.
+ */
+
+import test from "node:test";
+import assert from "node:assert/strict";
+import fs from "node:fs";
+import os from "node:os";
+import path from "node:path";
+
+const TEST_DATA_DIR = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-chat-synced-11089-"));
+process.env.DATA_DIR = TEST_DATA_DIR;
+
+const core = await import("../../src/lib/db/core.ts");
+const providersDb = await import("../../src/lib/db/providers.ts");
+const modelsDb = await import("../../src/lib/db/models.ts");
+const auth = await import("../../src/sse/services/auth.ts");
+
+const PROVIDER = "ollama-local";
+
+async function resetStorage() {
+ core.resetDbInstance();
+ fs.rmSync(TEST_DATA_DIR, { recursive: true, force: true });
+ fs.mkdirSync(TEST_DATA_DIR, { recursive: true });
+}
+
+test.after(() => {
+ core.resetDbInstance();
+ fs.rmSync(TEST_DATA_DIR, { recursive: true, force: true });
+});
+
+async function createConnection(data: Record): Promise {
+ const created = (await providersDb.createProviderConnection(data)) as { id: string };
+ return created.id;
+}
+
+/** The connection id the selector handed back, or null if it returned no account. */
+function selectedConnectionId(selected: unknown): string | null {
+ if (!selected || typeof selected !== "object") return null;
+ const id = (selected as { connectionId?: unknown }).connectionId;
+ return typeof id === "string" ? id : null;
+}
+
+/** Create the two-host ollama-local topology from the issue report. */
+async function seedTwoHosts(options: { studioRateLimitedUntil?: string } = {}) {
+ const studioId = await createConnection({
+ provider: PROVIDER,
+ authType: "none",
+ name: "Mac Studio",
+ baseUrl: "http://studio.lan:11434/v1",
+ priority: 1,
+ isActive: true,
+ });
+ const jetsonId = await createConnection({
+ provider: PROVIDER,
+ authType: "none",
+ name: "Jetson",
+ baseUrl: "http://jetson.lan:11434/v1",
+ priority: 2,
+ isActive: true,
+ });
+
+ await modelsDb.replaceSyncedAvailableModelsForConnection(PROVIDER, studioId, [
+ { id: "gemma3:4b", name: "gemma3:4b" },
+ { id: "flux2-klein:9b", name: "flux2-klein:9b" },
+ ]);
+ await modelsDb.replaceSyncedAvailableModelsForConnection(PROVIDER, jetsonId, [
+ { id: "gemma3:4b", name: "gemma3:4b" },
+ ]);
+
+ if (options.studioRateLimitedUntil) {
+ await providersDb.updateProviderConnection(studioId, {
+ rateLimitedUntil: options.studioRateLimitedUntil,
+ });
+ }
+
+ return { studioId, jetsonId };
+}
+
+test("#11089 selects only the host whose synced inventory advertises the model", async () => {
+ await resetStorage();
+ const { studioId, jetsonId } = await seedTwoHosts();
+
+ // Exclude studio to force the selector to look elsewhere. Jetson does not
+ // advertise flux2-klein:9b, so it must NOT be handed back.
+ const selected = await auth.getProviderCredentials(PROVIDER, studioId, null, "flux2-klein:9b");
+
+ assert.notEqual(
+ selectedConnectionId(selected),
+ jetsonId,
+ "jetson never synced flux2-klein:9b and must not be selected for it"
+ );
+});
+
+test("#11089 does not preemptively fail over to a host lacking the model when the owner is cooling", async () => {
+ await resetStorage();
+ const coolingUntil = new Date(Date.now() + 10 * 60 * 1000).toISOString();
+ const { jetsonId } = await seedTwoHosts({ studioRateLimitedUntil: coolingUntil });
+
+ const selected = await auth.getProviderCredentials(PROVIDER, null, null, "flux2-klein:9b");
+
+ assert.notEqual(
+ selectedConnectionId(selected),
+ jetsonId,
+ "a cooling studio must surface a cooldown, not silently route to a host without the model"
+ );
+});
+
+test("#11089 keeps the connection that does advertise the model selectable", async () => {
+ await resetStorage();
+ const { studioId } = await seedTwoHosts();
+
+ const selected = await auth.getProviderCredentials(PROVIDER, null, null, "flux2-klein:9b");
+
+ assert.equal(
+ selectedConnectionId(selected),
+ studioId,
+ "studio advertises flux2-klein:9b and must be selected"
+ );
+});
+
+test("#11089 a model advertised by every host leaves both connections eligible", async () => {
+ await resetStorage();
+ const { studioId, jetsonId } = await seedTwoHosts();
+
+ const first = await auth.getProviderCredentials(PROVIDER, null, null, "gemma3:4b");
+ assert.equal(
+ selectedConnectionId(first),
+ studioId,
+ "fill-first prefers priority 1 for a shared model"
+ );
+
+ // Excluding studio (the normal account-fallback path) must still reach jetson,
+ // because jetson genuinely advertises gemma3:4b.
+ const second = await auth.getProviderCredentials(PROVIDER, studioId, null, "gemma3:4b");
+ assert.equal(
+ selectedConnectionId(second),
+ jetsonId,
+ "jetson advertises gemma3:4b and must remain a valid failover"
+ );
+});
+
+test("#11089 fails open when the provider has no synced inventory at all", async () => {
+ await resetStorage();
+
+ const connectionId = await createConnection({
+ provider: PROVIDER,
+ authType: "none",
+ baseUrl: "http://127.0.0.1:11434/v1",
+ priority: 1,
+ isActive: true,
+ });
+
+ // No replaceSyncedAvailableModelsForConnection call: discovery never ran.
+ // Routing must behave exactly as before rather than filtering everything out.
+ const selected = await auth.getProviderCredentials(PROVIDER, null, null, "never-synced-model");
+
+ assert.equal(
+ selectedConnectionId(selected),
+ connectionId,
+ "an unsynced provider must not be filtered to zero candidates"
+ );
+});
diff --git a/tests/unit/chatcore-translation-paths.test.ts b/tests/unit/chatcore-translation-paths.test.ts
index 08cc564cc4e..6ecb0487889 100644
--- a/tests/unit/chatcore-translation-paths.test.ts
+++ b/tests/unit/chatcore-translation-paths.test.ts
@@ -369,7 +369,7 @@ async function invokeChatCore({
onCredentialsRefreshed = null,
onRequestSuccess = null,
sessionAffinityKey = null,
- reasoningTransportFallback = "skip",
+ reasoningTransportFallback = "drop",
managedLease = null,
cachedSettings = null,
}: any = {}) {
@@ -631,7 +631,7 @@ test("chatCore translates a streaming Responses upstream for a Chat client", asy
assert.match(streamed, /"content":"ok"/);
assert.match(streamed, /data: \[DONE\]/);
});
-test("chatCore rejects opaque reasoning for unknown Responses targets unless explicitly enabled", async () => {
+test("chatCore drops opaque reasoning for plaintext Responses targets by default (#10959)", async () => {
const reasoningItems = [
{ id: "rs_valid", type: "reasoning", encrypted_content: "encrypted-blob" },
{ type: "reasoning", encrypted_content: "" },
@@ -640,7 +640,7 @@ test("chatCore rejects opaque reasoning for unknown Responses targets unless exp
{ id: "fc_call", type: "function_call", call_id: "call_1", name: "search", arguments: "{}" },
];
- const rejected = await invokeChatCore({
+ const dropped = await invokeChatCore({
provider: "openai-compatible-sp-openai",
model: "gpt-5.4",
endpoint: "/v1/responses",
@@ -656,9 +656,12 @@ test("chatCore rejects opaque reasoning for unknown Responses targets unless exp
responseFormat: "openai-responses",
});
- assert.equal(rejected.result.success, false);
- assert.equal(rejected.result.status, 400);
- assert.equal(rejected.calls.length, 0);
+ assert.equal(dropped.result.success, true);
+ assert.equal(dropped.calls.length, 1);
+ assert.deepEqual(
+ dropped.call.body.input.filter((item) => item.type === "reasoning"),
+ [{ type: "reasoning", summary: [{ text: "not self-contained" }] }]
+ );
const enabled = await invokeChatCore({
provider: "openai-compatible-sp-openai",
@@ -682,7 +685,7 @@ test("chatCore rejects opaque reasoning for unknown Responses targets unless exp
assert.deepEqual(
input.filter((item) => item.type === "reasoning"),
[
- { id: "rs_valid", type: "reasoning", encrypted_content: "encrypted-blob" },
+ { id: "rs_valid", type: "reasoning", encrypted_content: "encrypted-blob", summary: [] }, // summary defaulted by #11110
{ type: "reasoning", summary: [{ text: "not self-contained" }] },
]
);
@@ -693,9 +696,9 @@ test("chatCore rejects opaque reasoning for unknown Responses targets unless exp
assert.equal(input.find((item) => item.type === "function_call")?.id, undefined);
});
-test("chatCore applies Chat reasoning compatibility before stream mode diverges", async () => {
+test("chatCore drops incompatible Chat reasoning before stream mode diverges (#10959)", async () => {
for (const stream of [false, true]) {
- const rejected = await invokeChatCore({
+ const dropped = await invokeChatCore({
provider: "openai-compatible-sp-openai",
model: "gpt-5.4",
endpoint: "/v1/chat/completions",
@@ -728,14 +731,14 @@ test("chatCore applies Chat reasoning compatibility before stream mode diverges"
},
});
- assert.equal(rejected.result.success, false, `stream=${stream}`);
- assert.equal(rejected.result.status, 400, `stream=${stream}`);
- assert.equal(rejected.calls.length, 0, `stream=${stream}`);
+ assert.equal(dropped.result.success, true, `stream=${stream}`);
+ assert.equal(dropped.calls.length, 1, `stream=${stream}`);
+ assert.equal(dropped.call.body.messages[0].reasoning_details, undefined, `stream=${stream}`);
}
});
-test("chatCore can drop incompatible reasoning for an opted-in Combo attempt", async () => {
- const dropped = await invokeChatCore({
+test("chatCore preserves Combo skip behavior for incompatible reasoning", async () => {
+ const skipped = await invokeChatCore({
provider: "openai-compatible-sp-openai",
model: "gpt-5.4",
endpoint: "/v1/responses",
@@ -757,15 +760,12 @@ test("chatCore can drop incompatible reasoning for an opted-in Combo attempt", a
},
responseFormat: "openai-responses",
isCombo: true,
- reasoningTransportFallback: "drop",
+ reasoningTransportFallback: "skip",
});
- assert.equal(dropped.result.success, true);
- assert.equal(dropped.calls.length, 1);
- assert.equal(
- dropped.call.body.input.some((item) => item.type === "reasoning"),
- false
- );
+ assert.equal(skipped.result.success, false);
+ assert.equal(skipped.result.status, 400);
+ assert.equal(skipped.calls.length, 0);
});
test("chatCore carries Chat reasoning_content into official DeepSeek Responses input", async () => {
@@ -800,6 +800,7 @@ test("chatCore carries Chat reasoning_content into official DeepSeek Responses i
assert.deepEqual(call.body.input.slice(0, 3), [
{
type: "reasoning",
+ summary: [], // defaulted on freshly-built reasoning items (#11129)
content: [{ type: "reasoning_text", text: "Inspect before calling the tool" }],
},
{
@@ -866,7 +867,7 @@ test("chatCore replays nonstream DeepSeek Responses reasoning across a Chat tool
assert.equal(second.result.success, true);
assert.deepEqual(
second.call.body.input.find((item) => item.type === "reasoning"),
- { type: "reasoning", content: [{ type: "reasoning_text", text: reasoning }] }
+ { type: "reasoning", content: [{ type: "reasoning_text", text: reasoning }], summary: [] } // summary defaulted by #11129
);
});
@@ -930,7 +931,7 @@ test("chatCore replays streamed DeepSeek Responses reasoning across a Chat tool
assert.equal(second.result.success, true);
assert.deepEqual(
second.call.body.input.find((item) => item.type === "reasoning"),
- { type: "reasoning", content: [{ type: "reasoning_text", text: reasoning }] }
+ { type: "reasoning", content: [{ type: "reasoning_text", text: reasoning }], summary: [] } // summary defaulted by #11129
);
});
@@ -1132,7 +1133,7 @@ test("chatCore automatically preserves provider-generated opaque reasoning for C
assert.equal(result.success, true);
assert.deepEqual(
call.body.input.filter((item) => item.type === "reasoning"),
- [{ id: "rs_valid", type: "reasoning", encrypted_content: "encrypted-blob" }]
+ [{ id: "rs_valid", type: "reasoning", encrypted_content: "encrypted-blob", summary: [] }] // summary defaulted by #11110
);
assert.equal(
call.body.input.some((item) => item.type === "item_reference"),
diff --git a/tests/unit/cli-combo-create-models-10954.test.ts b/tests/unit/cli-combo-create-models-10954.test.ts
index dc53459d6ed..52fabf8c169 100644
--- a/tests/unit/cli-combo-create-models-10954.test.ts
+++ b/tests/unit/cli-combo-create-models-10954.test.ts
@@ -90,7 +90,15 @@ test("combo create — parses --models without throwing (Commander option regist
});
await prog.parseAsync(
- ["node", "x", "combo", "create", "my-combo", "--models", "openai/gpt-4o,anthropic/claude-3-opus"],
+ [
+ "node",
+ "x",
+ "combo",
+ "create",
+ "my-combo",
+ "--models",
+ "openai/gpt-4o,anthropic/claude-3-opus",
+ ],
{ from: "node" }
);
@@ -222,3 +230,26 @@ test("combo create (HTTP) — POST /api/combos body carries the parsed models",
else process.env.DATA_DIR = ORIGINAL_DATA_DIR;
}
});
+
+// Regression for the follow-up of #11011: with --models available, creating
+// an empty combo is no longer a legitimate path on either transport.
+test("combo create without any model is refused before reaching a transport", async () => {
+ await withComboEnv(async () => {
+ const errors: string[] = [];
+ const originalError = console.error;
+ console.error = (msg?: unknown) => {
+ errors.push(String(msg));
+ };
+ try {
+ const mod = await import("../../bin/cli/commands/combo.mjs");
+ const rc = await mod.runComboCreateCommand("guard-test");
+ assert.equal(rc, 1);
+ } finally {
+ console.error = originalError;
+ }
+ assert.ok(
+ errors.some((m) => m.includes("--models")),
+ `stderr should name --models, got: ${errors.join(" | ")}`
+ );
+ });
+});
diff --git a/tests/unit/cli-helper/config-generator.test.ts b/tests/unit/cli-helper/config-generator.test.ts
index 1b2e85732fe..cfe3f398096 100644
--- a/tests/unit/cli-helper/config-generator.test.ts
+++ b/tests/unit/cli-helper/config-generator.test.ts
@@ -414,7 +414,7 @@ describe("config-generator", () => {
}
});
- it("does NOT fabricate a default context when the catalog has no entry", async () => {
+ it("uses the required 128K context fallback when the catalog has no entry", async () => {
const stub = stubFetchOnce(makeCatalogResponse(SAMPLE_CATALOG));
try {
const { generateOpencodeConfig } =
@@ -424,15 +424,14 @@ describe("config-generator", () => {
apiKey: "sk-test",
});
const cfg = JSON.parse(out);
- // NO_CTX_COMBO has no context_length in the catalog — generator
- // must NOT default to 128K (or any other value). The entry is
- // emitted without limit.context so OpenCode's own heuristic
- // applies and the user can fix the upstream.
+ // NO_CTX_COMBO has no context_length in the catalog. OpenCode v1
+ // requires a complete limit object, so the compatibility fallback
+ // must be explicit rather than leaving the config invalid.
const noCtx = cfg.provider.omniroute.models["NO_CTX_COMBO"];
assert.strictEqual(
noCtx.limit?.context,
- undefined,
- `NO_CTX_COMBO should not have a fabricated limit.context (got ${noCtx.limit?.context})`
+ 128_000,
+ `NO_CTX_COMBO should use the 128K fallback (got ${noCtx.limit?.context})`
);
} finally {
stub.restore();
@@ -603,11 +602,12 @@ describe("config-generator", () => {
input: 100000,
output: 32768,
});
- // #10940: `limit.output` is REQUIRED by OpenCode's v1 provider schema,
- // so even a model with zero catalog metadata still gets a `limit`
- // block carrying the fallback output value; `context`/`input` stay
- // omitted since neither the catalog nor the user knows them.
- assert.deepStrictEqual(models["no-metadata"].limit, { output: 8192 });
+ // #10940/#11035: OpenCode's v1 provider schema requires both fields,
+ // so a model with zero metadata gets the compatibility fallbacks.
+ assert.deepStrictEqual(models["no-metadata"].limit, {
+ context: 128_000,
+ output: 8192,
+ });
for (const model of Object.values(models) as Array<{ limit?: { output?: number } }>) {
assert.ok(
diff --git a/tests/unit/codex-drop-nonstandard-events.test.ts b/tests/unit/codex-drop-nonstandard-events.test.ts
index d2aff548149..dec24970ca5 100644
--- a/tests/unit/codex-drop-nonstandard-events.test.ts
+++ b/tests/unit/codex-drop-nonstandard-events.test.ts
@@ -17,6 +17,19 @@ function sseResponse(body: string): Response {
});
}
+function chunkedSseResponse(chunks: string[]): Response {
+ const encoder = new TextEncoder();
+ return new Response(
+ new ReadableStream({
+ start(controller) {
+ for (const chunk of chunks) controller.enqueue(encoder.encode(chunk));
+ controller.close();
+ },
+ }),
+ { status: 200, headers: { "content-type": "text/event-stream" } }
+ );
+}
+
async function readAll(res: Response): Promise {
return await res.text();
}
@@ -61,10 +74,10 @@ describe("codexDropNonstandardEvents (#11014)", () => {
describe("filterNonstandardCodexSse (#4715)", () => {
it("drops codex.* event blocks but keeps standard response.* events", async () => {
const stream =
- "event: response.created\ndata: {\"type\":\"response.created\"}\n\n" +
+ 'event: response.created\ndata: {"type":"response.created"}\n\n' +
"event: codex.rate_limits\n\n" +
- "event: response.output_text.delta\ndata: {\"delta\":\"hi\"}\n\n" +
- "event: response.completed\ndata: {\"type\":\"response.completed\"}\n\n";
+ 'event: response.output_text.delta\ndata: {"delta":"hi"}\n\n' +
+ 'event: response.completed\ndata: {"type":"response.completed"}\n\n';
const out = await readAll(filterNonstandardCodexSse(sseResponse(stream)));
assert.ok(!out.includes("codex.rate_limits"), "codex.* frame must be stripped");
assert.ok(out.includes("response.created"), "standard events preserved");
@@ -72,18 +85,31 @@ describe("filterNonstandardCodexSse (#4715)", () => {
assert.ok(out.includes("response.completed"), "terminal event preserved");
});
+ it("filters CRLF-framed events split across transport chunks", async () => {
+ const response = chunkedSseResponse([
+ 'event: response.created\r\ndata: {"type":"response.created"}\r\n\r',
+ "\nevent: codex.rate_limits\r\n\r\n",
+ 'event: response.completed\r\ndata: {"type":"response.completed"}\r\n\r\n',
+ ]);
+
+ const out = await readAll(filterNonstandardCodexSse(response));
+
+ assert.ok(!out.includes("codex.rate_limits"), "codex.* frame must be stripped");
+ assert.ok(out.includes("response.created"), "standard events preserved");
+ assert.ok(out.includes("response.completed"), "terminal event preserved");
+ });
+
it("passes through non-SSE responses untouched", async () => {
- const json = new Response("{\"ok\":true}", {
+ const json = new Response('{"ok":true}', {
status: 200,
headers: { "content-type": "application/json" },
});
const out = filterNonstandardCodexSse(json);
- assert.equal(await out.text(), "{\"ok\":true}");
+ assert.equal(await out.text(), '{"ok":true}');
});
it("drops a trailing codex.* block with no double-newline terminator (flush path)", async () => {
- const stream =
- "event: response.created\ndata: {}\n\n" + "event: codex.token_count\ndata: {}";
+ const stream = "event: response.created\ndata: {}\n\n" + "event: codex.token_count\ndata: {}";
const out = await readAll(filterNonstandardCodexSse(sseResponse(stream)));
assert.ok(out.includes("response.created"));
assert.ok(!out.includes("codex.token_count"));
diff --git a/tests/unit/combo-bracket-names.test.ts b/tests/unit/combo-bracket-names.test.ts
index 32844173cc1..0f4d8af0de9 100644
--- a/tests/unit/combo-bracket-names.test.ts
+++ b/tests/unit/combo-bracket-names.test.ts
@@ -30,6 +30,7 @@ test.after(() => {
test("combo schemas accept names with spaces and square brackets", () => {
const createResult = schemas.createComboSchema.safeParse({
name: "Claude [1m]",
+ models: ["anthropic/claude-3-opus"],
});
const updateResult = schemas.updateComboSchema.safeParse({
name: "Claude [1m]",
diff --git a/tests/unit/combo-context-length.test.ts b/tests/unit/combo-context-length.test.ts
index f47539d64ad..c416b4d8be3 100644
--- a/tests/unit/combo-context-length.test.ts
+++ b/tests/unit/combo-context-length.test.ts
@@ -46,6 +46,7 @@ test.after(async () => {
test("createComboSchema accepts valid context_length", () => {
const result = schemas.createComboSchema.safeParse({
name: "TestCombo",
+ models: ["openai/gpt-4o-mini"],
context_length: 128000,
});
assert.equal(result.success, true);
@@ -54,6 +55,7 @@ test("createComboSchema accepts valid context_length", () => {
test("createComboSchema rejects context_length below minimum (1000)", () => {
const result = schemas.createComboSchema.safeParse({
name: "TestCombo",
+ models: ["openai/gpt-4o-mini"],
context_length: 999,
});
assert.equal(result.success, false);
@@ -62,6 +64,7 @@ test("createComboSchema rejects context_length below minimum (1000)", () => {
test("createComboSchema rejects context_length above maximum (2000000)", () => {
const result = schemas.createComboSchema.safeParse({
name: "TestCombo",
+ models: ["openai/gpt-4o-mini"],
context_length: 2000001,
});
assert.equal(result.success, false);
@@ -70,12 +73,14 @@ test("createComboSchema rejects context_length above maximum (2000000)", () => {
test("createComboSchema accepts context_length at exact boundaries", () => {
const min = schemas.createComboSchema.safeParse({
name: "MinCombo",
+ models: ["openai/gpt-4o-mini"],
context_length: 1000,
});
assert.equal(min.success, true);
const max = schemas.createComboSchema.safeParse({
name: "MaxCombo",
+ models: ["openai/gpt-4o-mini"],
context_length: 2000000,
});
assert.equal(max.success, true);
@@ -84,6 +89,7 @@ test("createComboSchema accepts context_length at exact boundaries", () => {
test("createComboSchema rejects non-integer context_length", () => {
const result = schemas.createComboSchema.safeParse({
name: "TestCombo",
+ models: ["openai/gpt-4o-mini"],
context_length: 128000.5,
});
assert.equal(result.success, false);
@@ -92,6 +98,7 @@ test("createComboSchema rejects non-integer context_length", () => {
test("createComboSchema accepts omitted context_length", () => {
const result = schemas.createComboSchema.safeParse({
name: "TestCombo",
+ models: ["openai/gpt-4o-mini"],
});
assert.equal(result.success, true);
});
diff --git a/tests/unit/combo-empty-models.test.ts b/tests/unit/combo-empty-models.test.ts
index 1486628ec53..95f32813a34 100644
--- a/tests/unit/combo-empty-models.test.ts
+++ b/tests/unit/combo-empty-models.test.ts
@@ -24,9 +24,9 @@ test("an update cannot remove every model from a combo", () => {
assert.equal(updateComboSchema.safeParse({ name: "renamed" }).success, true);
});
-test("creating a combo with no model stays allowed — the CLI does it on purpose", () => {
- assert.equal(createComboSchema.safeParse({ name: "drafted", models: [] }).success, true);
- assert.equal(createComboSchema.safeParse({ name: "drafted" }).success, true);
+test("creating a combo without a model is refused at the boundary", () => {
+ assert.equal(createComboSchema.safeParse({ name: "drafted", models: [] }).success, false);
+ assert.equal(createComboSchema.safeParse({ name: "drafted" }).success, false);
});
test("the copilot createCombo tool stores targets where the router looks for them", async () => {
diff --git a/tests/unit/combo-quota-exhaustion-only-fallback.test.ts b/tests/unit/combo-quota-exhaustion-only-fallback.test.ts
index 2a90d7e3fbd..cbb4f70c11f 100644
--- a/tests/unit/combo-quota-exhaustion-only-fallback.test.ts
+++ b/tests/unit/combo-quota-exhaustion-only-fallback.test.ts
@@ -110,7 +110,7 @@ async function run(
}
test("quota classifier rejects terminal-looking evidence on ineligible statuses", async () => {
- for (const status of [400, 401, 403, 404, 408, 409, 422, 500, 502, 503, 504]) {
+ for (const status of [400, 401, 404, 408, 409, 422, 500, 502, 503, 504]) {
for (const terminal of ["insufficient_quota", "quota_exhausted", "credits_exhausted"]) {
assert.equal(
await isQuotaExhaustionResponse(
diff --git a/tests/unit/combo-runtime-unit-concurrency.test.ts b/tests/unit/combo-runtime-unit-concurrency.test.ts
index 36a14a59ce7..8ccebe4603e 100644
--- a/tests/unit/combo-runtime-unit-concurrency.test.ts
+++ b/tests/unit/combo-runtime-unit-concurrency.test.ts
@@ -28,8 +28,8 @@ const databases = db.pragma("database_list") as Array<{ file?: string; name?: st
const activeDbPath = databases.find((database) => database.name === "main")?.file;
assert.ok(activeDbPath, "test requires a file-backed main SQLite database");
assert.equal(
- path.dirname(path.resolve(activeDbPath)),
- path.resolve(TEST_DATA_DIR),
+ fs.realpathSync(path.dirname(path.resolve(activeDbPath))),
+ fs.realpathSync(path.resolve(TEST_DATA_DIR)),
`active test database must be under TEST_DATA_DIR before inserts: ${activeDbPath}`
);
diff --git a/tests/unit/compression/ccr-non-mcp-full-prompt-loss-7746.test.ts b/tests/unit/compression/ccr-non-mcp-full-prompt-loss-7746.test.ts
index 0f4abd534e6..a646744871c 100644
--- a/tests/unit/compression/ccr-non-mcp-full-prompt-loss-7746.test.ts
+++ b/tests/unit/compression/ccr-non-mcp-full-prompt-loss-7746.test.ts
@@ -15,6 +15,7 @@ import { describe, it, before } from "node:test";
import assert from "node:assert/strict";
import {
ccrEngine,
+ getCcrStoreStats,
resetCcrStore,
retrieveBlock,
} from "../../../open-sse/services/compression/engines/ccr/index.ts";
@@ -54,39 +55,117 @@ describe("issue #7746 — CCR must not reduce the sole user prompt to a bare, un
});
it("prompt fixture is realistically sized (>= default 600-char minChars)", () => {
- assert.ok(REPORTER_PROMPT.length >= 600, `fixture must be >= 600 chars, got ${REPORTER_PROMPT.length}`);
+ assert.ok(
+ REPORTER_PROMPT.length >= 600,
+ `fixture must be >= 600 chars, got ${REPORTER_PROMPT.length}`
+ );
});
- it("does not leave the model with only the bare CCR marker when no retrieve tool is available", () => {
+ it("non-MCP caller: CCR skips entirely — the sole user prompt passes through verbatim", () => {
resetCcrStore();
const body = makeOpenCodeStyleRequestBody();
const result = ccrEngine.apply(body as Record, { stepConfig: {} });
- assert.equal(result.compressed, true, "CCR compressed the sole user message (reproducing the report)");
-
- const messages = result.body.messages as Array<{ role: string; content: string }>;
- const compressedContent = messages[0].content;
- const isBareMarkerOnly = /^\[CCR retrieve hash=[0-9a-f]{24} chars=\d+\]$/.test(compressedContent);
-
+ // #7746 follow-up (forge review outage, 2026-08-22): the preamble guard was
+ // not enough — a non-MCP caller received "[CCR retrieve hash=...] markers"
+ // it had no tool to resolve (upstream saw 112 of ~3.6K tokens). The engine
+ // now refuses to replace content at all when tools[] lacks
+ // omniroute_ccr_retrieve: compressed=false, message content untouched.
assert.equal(
- isBareMarkerOnly,
+ result.compressed,
false,
- "BUG #7746: CCR replaced the ENTIRE sole user message with nothing but the bare " +
- `[CCR retrieve hash=...] marker, permanently losing the original prompt for any ` +
- `non-MCP caller that cannot resolve the marker. Got: ${JSON.stringify(compressedContent)}`
+ "CCR must not compress for a caller without the retrieve tool"
);
+ assert.equal(result.stats, null, "no stats when the engine is skipped");
+ const messages = result.body.messages as Array<{ role: string; content: string }>;
+ assert.equal(messages[0].role, "user", "message role must stay user");
+ assert.equal(
+ messages[0].content,
+ REPORTER_PROMPT,
+ "sole user prompt must pass through verbatim"
+ );
+ assert.equal(messages.length, 1, "no protocol instruction may be injected for non-MCP callers");
+ // Guard regression check: if callerSupportsCcrRetrieve ever returned true
+ // here, the store would silently accumulate blocks no non-MCP caller can
+ // retrieve. After a skip the store must hold nothing for this principal.
+ assert.equal(getCcrStoreStats().entries, 0, "store must stay empty after a non-MCP skip");
});
- it("the original prompt remains fully retrievable by hash even after the guard applies", () => {
+ // tools:[] and unrelated tools are distinct caller shapes that must all be
+ // treated as non-MCP: an empty array and a foreign tool list both mean the
+ // retrieve tool is unreachable.
+ for (const label of ["empty tools array", "unrelated tools"] as const) {
+ it(`non-MCP caller with ${label}: CCR skips entirely`, () => {
+ resetCcrStore();
+ const tools =
+ label === "empty tools array"
+ ? []
+ : [
+ { type: "function", function: { name: "get_weather" } },
+ { type: "function", function: { name: "web_search" } },
+ ];
+ const body = { ...makeOpenCodeStyleRequestBody(), tools };
+ const result = ccrEngine.apply(body as Record, { stepConfig: {} });
+
+ assert.equal(result.compressed, false, `${label} must not compress`);
+ const messages = result.body.messages as Array<{ role: string; content: string }>;
+ assert.equal(messages[0].content, REPORTER_PROMPT, "prompt passes through verbatim");
+ assert.equal(messages.length, 1, "no protocol instruction injected");
+ });
+ }
+
+ // A malformed body (tools as a non-array, or entries of unexpected shape)
+ // must fail OPEN — no compression, never a throw into the request pipeline.
+ for (const malformed of [
+ { tools: "not-an-array" },
+ { tools: [null, 42, "x"] },
+ { tools: [{}, { type: "function" }] },
+ ]) {
+ it(`malformed tools payload (${JSON.stringify(malformed.tools)}): engine skips without throwing`, () => {
+ resetCcrStore();
+ const body = { ...makeOpenCodeStyleRequestBody(), ...malformed };
+ const result = ccrEngine.apply(body as Record, { stepConfig: {} });
+
+ assert.equal(result.compressed, false, "malformed tools must fail open (skip)");
+ const messages = result.body.messages as Array<{ role: string; content: string }>;
+ assert.equal(messages[0].content, REPORTER_PROMPT, "prompt passes through verbatim");
+ assert.equal(
+ getCcrStoreStats().entries,
+ 0,
+ "store must stay empty after a malformed-tools skip"
+ );
+ });
+ }
+
+ it("MCP-capable caller (tools[] advertises omniroute_ccr_retrieve): replacement still runs and stays retrievable", () => {
resetCcrStore();
- const body = makeOpenCodeStyleRequestBody();
+ const body = {
+ ...makeOpenCodeStyleRequestBody(),
+ tools: [{ type: "function", function: { name: "omniroute_ccr_retrieve" } }],
+ };
const result = ccrEngine.apply(body as Record, { stepConfig: {} });
+ assert.equal(result.compressed, true, "CCR still compresses for MCP-capable callers");
const messages = result.body.messages as Array<{ role: string; content: string }>;
- const compressedContent = messages[0].content;
+ // The protocol instruction is injected as a leading system message, so the
+ // compressed conversation is exactly: [instruction, original user message].
+ assert.equal(messages.length, 2, "instruction + user message");
+ assert.equal(messages[0].role, "system", "instruction is a leading system message");
+ assert.ok(
+ typeof messages[0].content === "string" && messages[0].content.length > 0,
+ "instruction content must be non-empty"
+ );
+ assert.ok(
+ messages[0].content.includes("omniroute_ccr_retrieve"),
+ "instruction must teach the retrieve tool contract"
+ );
+ const compressedContent = messages[1].content;
const match = compressedContent.match(/\[CCR retrieve hash=([0-9a-f]{24}) chars=\d+\]/);
- assert.ok(match, "compressed content must still contain a resolvable CCR marker");
- const hash = match![1];
- assert.equal(retrieveBlock(hash), REPORTER_PROMPT, "original prompt must be stored verbatim and retrievable");
+ assert.ok(match, "compressed content must contain a resolvable CCR marker");
+ assert.equal(
+ retrieveBlock(match![1]),
+ REPORTER_PROMPT,
+ "original prompt must be stored verbatim and retrievable"
+ );
});
});
diff --git a/tests/unit/compression/ccr-retrieval-ramp.test.ts b/tests/unit/compression/ccr-retrieval-ramp.test.ts
index 2b475e69bfb..6e87f7b7877 100644
--- a/tests/unit/compression/ccr-retrieval-ramp.test.ts
+++ b/tests/unit/compression/ccr-retrieval-ramp.test.ts
@@ -78,7 +78,13 @@ describe("ccrEngine.apply — retrieval-aware compression (H8)", () => {
const block = (len: number) => "x".repeat(len);
const run = (content: string, retrievalRampFactor = 2) =>
ccrEngine.apply(
- { messages: [{ role: "user", content }] },
+ // The retrieve tool is advertised — this suite exercises the compression
+ // path itself (H8 ramp); without the tool declaration the #7746 guard
+ // skips the engine entirely.
+ {
+ messages: [{ role: "user", content }],
+ tools: [{ type: "function", function: { name: "omniroute_ccr_retrieve" } }],
+ },
{ stepConfig: { minChars: BASE, retrievalRampFactor }, principalId: P }
);
diff --git a/tests/unit/context-manager-purify-system-first.test.ts b/tests/unit/context-manager-purify-system-first.test.ts
new file mode 100644
index 00000000000..51c658231d0
--- /dev/null
+++ b/tests/unit/context-manager-purify-system-first.test.ts
@@ -0,0 +1,91 @@
+import { test } from "node:test";
+import assert from "node:assert/strict";
+import { compressContext } from "../../open-sse/services/contextManager.ts";
+
+/**
+ * Plan-A root fix for the 2026-08-22 tokenrouter 400s. purifyHistory() used to
+ * splice the `[Context compressed: …]` notice as a SECOND system-role message at
+ * index system.length; strict gateways (TokenRouter, xiaomi-mimo/mimo) reject any
+ * system message at index > 0 with HTTP 400 "System message must be at the
+ * beginning". The notice must now merge into the leading system/developer
+ * message — or prepend a single system message when none exists — so the output
+ * never contains a system role after index 0, for ANY provider.
+ */
+
+function bigTurn(n: number) {
+ return { role: "user", content: `turn ${n}: ${"x".repeat(4_000)}` };
+}
+
+function run(body: Record) {
+ // ~30k tokens of history vs a small target forces Layer-3 purify_history.
+ return compressContext(body, { maxTokens: 5_000, reserveTokens: 0 });
+}
+
+function systemIndices(messages: Array<{ role: string }>) {
+ return messages.map((m, i) => (m.role === "system" ? i : -1)).filter((i) => i >= 0);
+}
+
+test("purify_history merges dropped-notice into existing leading system message", () => {
+ const body = {
+ model: "any-model",
+ messages: [
+ { role: "system", content: "You are a helpful assistant." },
+ ...Array.from({ length: 12 }, (_, i) => bigTurn(i)),
+ ],
+ };
+ const result = run(body);
+ assert.equal(result.compressed, true);
+ const messages = (result.body as { messages: Array> }).messages;
+ assert.deepEqual(systemIndices(messages as Array<{ role: string }>).slice(1), []);
+ const first = messages[0];
+ assert.equal(first.role, "system");
+ const text = String(first.content);
+ assert.match(text, /Context compressed: \d+ earlier messages removed/);
+ assert.match(text, /You are a helpful assistant\./);
+});
+
+test("purify_history prepends a single system notice when no system message exists", () => {
+ const body = {
+ model: "any-model",
+ messages: Array.from({ length: 12 }, (_, i) => bigTurn(i)),
+ };
+ const result = run(body);
+ assert.equal(result.compressed, true);
+ const messages = (result.body as { messages: Array> }).messages;
+ assert.deepEqual(systemIndices(messages as Array<{ role: string }>), [0]);
+ assert.match(String(messages[0].content), /Context compressed: \d+ earlier messages removed/);
+});
+
+test("purify_history merges into leading developer message without adding a second one", () => {
+ const body = {
+ model: "any-model",
+ messages: [
+ { role: "developer", content: "dev instructions" },
+ ...Array.from({ length: 12 }, (_, i) => bigTurn(i)),
+ ],
+ };
+ const result = run(body);
+ assert.equal(result.compressed, true);
+ const messages = (result.body as { messages: Array> }).messages;
+ assert.deepEqual(
+ messages.filter((m) => m.role === "developer").length,
+ 1,
+ "exactly one developer message"
+ );
+ assert.match(String(messages[0].content), /Context compressed: \d+ earlier messages removed/);
+ assert.match(String(messages[0].content), /dev instructions/);
+});
+
+test("no compression means no notice and untouched history", () => {
+ const body = {
+ model: "any-model",
+ messages: [
+ { role: "system", content: "sys" },
+ { role: "user", content: "hi" },
+ ],
+ };
+ const result = run(body);
+ assert.equal(result.compressed, false);
+ const messages = (result.body as { messages: unknown[] }).messages;
+ assert.equal(messages.length, 2);
+});
diff --git a/tests/unit/cursor-image-input.test.ts b/tests/unit/cursor-image-input.test.ts
index 16bffcb6b30..abc14d7d983 100644
--- a/tests/unit/cursor-image-input.test.ts
+++ b/tests/unit/cursor-image-input.test.ts
@@ -1,4 +1,4 @@
-import test from "node:test";
+import test, { type TestContext } from "node:test";
import assert from "node:assert/strict";
import crypto from "node:crypto";
import dns from "node:dns";
@@ -473,7 +473,37 @@ test("resolveCursorImages soft-caps a large PNG under the wire budget", async ()
// ─── Executor-level error body (response path, hard rule #12) ───────────────
-test("executor returns a sanitized 400 for an oversized image", async () => {
+// #10804 moved agent-endpoint discovery (a live api2.cursor.sh call) ahead of
+// request building inside CursorExecutor.execute. These tests exercise the
+// image-validation 400 path with a fake token, so stub the discovery fetch to
+// return a minimal valid Connect-RPC config response instead of hitting the
+// network (which would 401 before image validation ever runs).
+function mockCursorServerConfig(t: TestContext): void {
+ t.mock.method(globalThis, "fetch", async (input, init) => {
+ const url = String(input);
+ if (!url.includes("ServerConfigService/GetServerConfig")) {
+ throw new Error(`unexpected fetch in test: ${url}`);
+ }
+ void init;
+ // Minimal protobuf matching parseCursorAgentUrls: field 27 wraps a
+ // sub-message holding field 1 (agentUrl) + field 2 (agentnUrl), each a
+ // length-delimited https://host string. validateCursorAgentUrl only
+ // accepts *.api5.cursor.sh hosts, so use those.
+ const str = (field: number, host: string): Buffer => {
+ const value = Buffer.from(`https://${host}`);
+ return Buffer.concat([Buffer.from([(field << 3) | 0x02, value.length]), value]);
+ };
+ const inner = Buffer.concat([str(1, "us.api5.cursor.sh"), str(2, "eu.api5.cursor.sh")]);
+ // Field-27 tag (218) needs proper varint encoding (2 bytes).
+ const tag = ((27 << 3) | 0x02) as number;
+ const header = Buffer.from([(tag & 0x7f) | 0x80, tag >>> 7, inner.length]);
+ const body = Buffer.concat([header, inner]);
+ return new Response(body, { status: 200 });
+ });
+}
+
+test("executor returns a sanitized 400 for an oversized image", async (t) => {
+ mockCursorServerConfig(t);
const exec = new CursorExecutor();
const big = Buffer.alloc(MAX_CURSOR_IMAGE_DECODE_BYTES + 16).toString("base64");
const result = await exec.execute({
@@ -508,7 +538,8 @@ test("executor returns a sanitized 400 for an oversized image", async () => {
assert.ok(!/\/(root|home|usr)\//.test(body.error.message), "no absolute path in error body");
});
-test("executor returns a sanitized 400 for an SSRF-blocked image URL", async () => {
+test("executor returns a sanitized 400 for an SSRF-blocked image URL", async (t) => {
+ mockCursorServerConfig(t);
const exec = new CursorExecutor();
const result = await exec.execute({
model: "gpt-5.2",
diff --git a/tests/unit/dashboard-ux-operability.test.ts b/tests/unit/dashboard-ux-operability.test.ts
new file mode 100644
index 00000000000..cd3e3b533ad
--- /dev/null
+++ b/tests/unit/dashboard-ux-operability.test.ts
@@ -0,0 +1,9 @@
+import assert from "node:assert/strict";
+import { test } from "node:test";
+import { formatQuotaLabel } from "../../src/app/(dashboard)/dashboard/usage/components/ProviderLimits/utils.tsx";
+
+test("formatQuotaLabel formats custom quota keys with proper title-casing", () => {
+ assert.equal(formatQuotaLabel("session"), "Session");
+ assert.equal(formatQuotaLabel("weekly"), "Weekly");
+ assert.equal(formatQuotaLabel("custom_quota_limit"), "Custom Quota Limit");
+});
diff --git a/tests/unit/discontinued-providers-2026.test.ts b/tests/unit/discontinued-providers-2026.test.ts
index 1979a7cd9aa..88f963ef911 100644
--- a/tests/unit/discontinued-providers-2026.test.ts
+++ b/tests/unit/discontinued-providers-2026.test.ts
@@ -6,6 +6,9 @@ import assert from "node:assert";
// free tier that does not exist. The budget catalog already dropped them. The 2026-06-18 batch
// (gitlawb, gitlawb-gmi, aimlapi, yi) was each re-verified against the official source before flipping
// (aimlapi docs: "The Free Tier is currently paused"; gitlawb GitHub issue #1345: MiMo revoked).
+// 2026-08-22 (#10071): the five g4f.space sub-providers lost their anonymous tier to a proof-of-work
+// credit wall (keyless POST -> HTTP 402 insufficient_credits). They remain usable with a g4f.dev
+// member key, so only hasFree/freeNote/authHint changed - registry wiring is untouched.
describe("2026 discontinued free tiers — providers.ts hasFree reconciliation", () => {
it("APIKEY_PROVIDERS dead tiers no longer advertise a free tier", async () => {
const { APIKEY_PROVIDERS } = await import("../../src/shared/constants/providers.ts");
@@ -27,6 +30,42 @@ describe("2026 discontinued free tiers — providers.ts hasFree reconciliation",
}
});
+ it("g4f.space sub-providers no longer advertise an anonymous free tier", async () => {
+ const { APIKEY_PROVIDERS } = await import("../../src/shared/constants/providers.ts");
+ // 2026-08-22 live re-verification: every g4f.space sub-path still lists models keylessly, but a
+ // keyless POST /v1/chat/completions returns HTTP 402 {"type":"insufficient_credits"} pointing at
+ // a proof-of-work "cake" wall (g4f.dev/chat) or a member key (g4f.dev/members.html). The gateway
+ // is NOT dead - it works with a g4f.dev member key - so the registry entries and their
+ // authType:"optional" are deliberately untouched; only the free-tier advertisement is corrected.
+ for (const id of ["g4f-groq", "g4f-gemini", "g4f-pollinations", "g4f-ollama", "g4f-nvidia"]) {
+ const p = (
+ APIKEY_PROVIDERS as Record<
+ string,
+ { hasFree?: boolean; freeNote?: string; authHint?: string }
+ >
+ )[id];
+ assert.ok(
+ p,
+ `${id} should still exist in APIKEY_PROVIDERS (gateway still usable with a member key)`
+ );
+ assert.strictEqual(
+ p.hasFree,
+ false,
+ `${id} should have hasFree:false (anonymous tier walled behind proof-of-work credits in 2026)`
+ );
+ assert.match(
+ p.freeNote ?? "",
+ /proof-of-work/i,
+ `${id} freeNote should explain the proof-of-work credit wall`
+ );
+ assert.match(
+ p.authHint ?? "",
+ /member key/i,
+ `${id} authHint should state that a g4f.dev member key is required`
+ );
+ }
+ });
+
it("phind is fully removed (service shut down 2026-01) from both catalogs", async () => {
const { APIKEY_PROVIDERS, WEB_COOKIE_PROVIDERS } =
await import("../../src/shared/constants/providers.ts");
diff --git a/tests/unit/empty-stream-no-content-8649.test.ts b/tests/unit/empty-stream-no-content-8649.test.ts
index e46157fbe17..918ae7c4c06 100644
--- a/tests/unit/empty-stream-no-content-8649.test.ts
+++ b/tests/unit/empty-stream-no-content-8649.test.ts
@@ -252,3 +252,68 @@ test("#8649 buildStreamErrorChunks-shaped error must not be rewritten as empty c
assert.match(text, /AI Model Not Found/);
assert.doesNotMatch(text, /Provider returned empty content/);
});
+
+test("#8649 a Responses compaction-only stream is real output, not empty content", async () => {
+ // Codex remote compaction V2: POST /v1/responses with a compaction_trigger
+ // input item completes with output = [{type:"compaction", encrypted_content}]
+ // and no assistant text. The watcher's content keys do not include
+ // encrypted_content, so the healthy stream was followed by a synthetic
+ // response.failed ("Provider returned empty content") — which strict
+ // Responses clients reject even after response.completed.
+ const text = await runClientStream(
+ [
+ `data: {"type":"response.in_progress"}\n\n`,
+ `event: response.created\ndata: ${JSON.stringify({
+ type: "response.created",
+ response: { id: "resp_cmp", status: "in_progress", output: [] },
+ })}\n\n`,
+ `event: response.completed\ndata: ${JSON.stringify({
+ type: "response.completed",
+ response: {
+ id: "resp_cmp",
+ status: "completed",
+ output: [
+ { id: "cmp_1", type: "compaction", encrypted_content: "gAAAAABencryptedpayload" },
+ ],
+ },
+ })}\n\n`,
+ ],
+ FORMATS.OPENAI_RESPONSES
+ );
+
+ assert.match(text, /"type":"compaction"/);
+ assert.doesNotMatch(
+ text,
+ /Provider returned empty content|response\.failed/,
+ "a completed compaction response must not be followed by a synthetic failure frame"
+ );
+});
+
+test("#8649 an encrypted-reasoning-only stream is still empty content", async () => {
+ // Inverse of the compaction carve-out: an encrypted reasoning item is not
+ // user-visible output. A turn that produces only a reasoning trace and no
+ // message/tool call is the fake-success shape this guard exists to catch.
+ const text = await runClientStream(
+ [
+ `event: response.created\ndata: ${JSON.stringify({
+ type: "response.created",
+ response: { id: "resp_r", status: "in_progress", output: [] },
+ })}\n\n`,
+ `event: response.completed\ndata: ${JSON.stringify({
+ type: "response.completed",
+ response: {
+ id: "resp_r",
+ status: "completed",
+ output: [{ id: "rs_1", type: "reasoning", encrypted_content: "gAAAAABencryptedtrace" }],
+ },
+ })}\n\n`,
+ ],
+ FORMATS.OPENAI_RESPONSES
+ );
+
+ assert.match(
+ text,
+ /response\.failed|Provider returned empty content/,
+ "a reasoning-only turn must keep tripping the empty-content guard"
+ );
+});
diff --git a/tests/unit/flat-rate-cost-5552.test.ts b/tests/unit/flat-rate-cost-5552.test.ts
index 2df378ee3e5..61ad9847b16 100644
--- a/tests/unit/flat-rate-cost-5552.test.ts
+++ b/tests/unit/flat-rate-cost-5552.test.ts
@@ -25,6 +25,7 @@ test("isFlatRateProvider: dedicated subscription / coding-plan providers are fla
"glm-cn",
"claude",
"cc",
+ "opencode-go",
]) {
assert.equal(isFlatRateProvider(id), true, `${id} should be flat-rate`);
}
@@ -83,6 +84,27 @@ test("computeCostFromPricing: opt-in only — flat-rate provider WITHOUT the fla
assert.equal(computeCostFromPricing(PRICING, TOKENS, { provider: "chatgpt-web" }), 3);
});
+test("#11149: opencode-go is a flat-rate subscription, not metered", () => {
+ // opencode-go (https://opencode.ai/go) is a $10/month flat subscription that
+ // resells GLM, Kimi, Grok, DeepSeek, MiniMax, Qwen and GPT-5.x. Because it is
+ // an aggregator, every call was priced at the UNDERLYING model's metered rate,
+ // so the overstatement is large rather than marginal (a reported ~$13.35 for a
+ // month actually billed at $10 flat). It is api-key auth, so it is not covered
+ // by the dynamic WEB_COOKIE_PROVIDERS branch and needs the explicit id.
+ assert.equal(isFlatRateProvider("opencode-go"), true);
+ assert.equal(
+ computeCostFromPricing(PRICING, TOKENS, { provider: "opencode-go", flatRateAsZero: true }),
+ 0
+ );
+ // Still opt-in: without the flag the per-request estimate is unchanged.
+ assert.equal(computeCostFromPricing(PRICING, TOKENS, { provider: "opencode-go" }), 3);
+});
+
+test("#11149: sibling opencode ids keep their own billing semantics", () => {
+ // Only the Go subscription is flat-rate. The keyless `opencode` provider is a
+ // different id and must not be swept in by a prefix-style match.
+ assert.equal(isFlatRateProvider("opencode"), false);
+});
test("computeCostFromPricing: metered provider with the flag still estimates", () => {
assert.equal(
computeCostFromPricing(PRICING, TOKENS, { provider: "openai", flatRateAsZero: true }),
diff --git a/tests/unit/g4f-space-gateway-6650.test.ts b/tests/unit/g4f-space-gateway-6650.test.ts
index 8f1aadc2bcd..93fd99bd679 100644
--- a/tests/unit/g4f-space-gateway-6650.test.ts
+++ b/tests/unit/g4f-space-gateway-6650.test.ts
@@ -17,7 +17,8 @@
* category on the dashboard
* - allowed to skip API key validation (providerAllowsOptionalApiKey)
* - has provider metadata (name/website/free-tier note) in the apikey
- * gateway catalog
+ * gateway catalog (hasFree flipped false by #10071 — anonymous tier now
+ * requires proof-of-work credits; a g4f.dev member key is required)
*/
import test from "node:test";
import assert from "node:assert/strict";
@@ -96,7 +97,10 @@ for (const [id, subPath] of Object.entries(SUB_PATHS)) {
assert.ok(meta, `${id} should have an APIKEY_PROVIDERS metadata entry`);
assert.equal(meta.id, id);
assert.equal(meta.website, "https://g4f.space");
- assert.equal(meta.hasFree, true);
+ // hasFree was true at #6650 time; the anonymous tier was walled behind proof-of-work
+ // credits in 2026 (#10071), so the flag is now false. Registry wiring above is unchanged:
+ // the provider still works with a g4f.dev member key, hence authType stays "optional".
+ assert.equal(meta.hasFree, false);
assert.equal(typeof meta.freeNote, "string");
assert.ok((meta.freeNote as string).length > 0);
});
diff --git a/tests/unit/glm-5.3-catalog-and-effort-tiers.test.ts b/tests/unit/glm-5.3-catalog-and-effort-tiers.test.ts
index d14c136e7b3..5d927de02a8 100644
--- a/tests/unit/glm-5.3-catalog-and-effort-tiers.test.ts
+++ b/tests/unit/glm-5.3-catalog-and-effort-tiers.test.ts
@@ -1,7 +1,7 @@
import test from "node:test";
import assert from "node:assert/strict";
-// GLM-5.3 support (released 2026-08-14, https://z.ai/blog/glm-5.3).
+// GLM-5.3 support (released 2026-08-14, https://docs.z.ai/guides/llm/glm-5.3).
//
// Upstream ships ONE model id (`glm-5.3`) — effort is a request parameter
// (`reasoning_effort`: low|high|max, default max) on the coding chat/completions
@@ -12,14 +12,15 @@ import assert from "node:assert/strict";
// beta header), the 5.3 tiers use the documented `reasoning_effort` param on the
// OpenAI coding transport.
//
-// Spec caveat: Z.ai has not yet published the default context window — 1M is
-// mirrored from GLM-5.2 (same base model) per operator decision; correct when
-// the official spec lands.
+// Z.AI documents a 1M context window and 128K maximum output.
-const { getRegistryEntry } = await import("../../open-sse/config/providerRegistry.ts");
+const { getRegistryEntry, REGISTRY } = await import("../../open-sse/config/providerRegistry.ts");
const { GlmExecutor } = await import("../../open-sse/executors/glm.ts");
const { MODEL_SPECS } = await import("../../src/shared/constants/modelSpecs.ts");
const { GLM_PRICING } = await import("../../src/shared/constants/pricing/shared-tiers.ts");
+const metadataRegistry = await import("../../src/lib/modelMetadataRegistry.ts");
+const { shouldExposeSyncedEffortVariants, SYNCED_EFFORT_SKIP_PROVIDERS } =
+ await import("../../open-sse/utils/syncedEffortVariants.ts");
const GLM_5_3_IDS = ["glm-5.3", "glm-5.3-high", "glm-5.3-low"] as const;
@@ -38,6 +39,87 @@ function modelIds(provider: string): string[] {
return (entry.models ?? []).map((m) => m.id);
}
+test("shared GLM providers keep their dedicated aliases instead of synthesizing another layer", () => {
+ for (const provider of ["glm", "glm-cn", "glmt"]) {
+ assert.ok(SYNCED_EFFORT_SKIP_PROVIDERS.has(provider), provider);
+ assert.equal(
+ shouldExposeSyncedEffortVariants({
+ id: `${provider}/glm-5.3`,
+ owned_by: provider,
+ capabilities: { effort_tiers: ["low", "high", "max"] },
+ }),
+ false,
+ provider
+ );
+ }
+ assert.equal(SYNCED_EFFORT_SKIP_PROVIDERS.has("zcode"), false);
+});
+
+test("GLM family detection covers numeric, Z1, and bare provider model ids", () => {
+ for (const modelId of [
+ "hf:zai-org/GLM-5.2",
+ "THUDM/GLM-Z1-32B-0414",
+ "THUDM/GLM-Z1-9B-0414",
+ "glm",
+ ]) {
+ assert.equal(metadataRegistry.isGlmFamilyModel(modelId), true, modelId);
+ }
+ assert.equal(metadataRegistry.isGlmFamilyModel("llama-3.3"), false);
+});
+
+test("catalog suppresses inferred tiers for every GLM registry entry without a provider contract", () => {
+ let audited = 0;
+ for (const [provider, entry] of Object.entries(REGISTRY)) {
+ for (const model of entry.models ?? []) {
+ if (!metadataRegistry.isGlmFamilyModel(model.id, model.name)) continue;
+ audited += 1;
+ const enriched = metadataRegistry.enrichCatalogModelEntry({
+ id: `${provider}/${model.id}`,
+ object: "model",
+ owned_by: provider,
+ root: model.id,
+ }) as Record;
+ const capabilities = enriched.capabilities as Record;
+ if (capabilities.supportsThinking === true) {
+ assert.deepEqual(
+ capabilities.effort_tiers,
+ model.supportedThinkingEfforts ?? [],
+ `${provider}/${model.id}`
+ );
+ } else {
+ assert.equal("effort_tiers" in capabilities, false, `${provider}/${model.id}`);
+ }
+ }
+ }
+ assert.ok(audited > 0);
+});
+
+test("catalog exposes only GLM effort tiers that each provider can route", () => {
+ const routedTiers = new Map([
+ ["glm-5.3", ["low", "high", "max"]],
+ ["glm-5.3-high", ["high"]],
+ ["glm-5.3-low", ["low"]],
+ ["glm-5.2", ["high", "max"]],
+ ["glm-5.2-high", ["high"]],
+ ["glm-5.2-max", ["max"]],
+ ]);
+
+ for (const provider of ["glm", "glm-cn", "glmt", "zcode"]) {
+ for (const model of getRegistryEntry(provider)!.models ?? []) {
+ const enriched = metadataRegistry.enrichCatalogModelEntry({
+ id: `${provider}/${model.id}`,
+ object: "model",
+ owned_by: provider,
+ root: model.id,
+ }) as Record;
+ const capabilities = enriched.capabilities as Record;
+ const expected = provider === "zcode" ? [] : (routedTiers.get(model.id) ?? []);
+ assert.equal(capabilities.supportsThinking, true, `${provider}/${model.id}`);
+ assert.deepEqual(capabilities.effort_tiers, expected, `${provider}/${model.id}`);
+ }
+ }
+});
+
for (const provider of ["glm", "glm-cn", "glmt"]) {
test(`${provider} advertises the GLM-5.3 base model and effort tiers (GLM_SHARED_MODELS)`, () => {
const ids = modelIds(provider);
diff --git a/tests/unit/glm-executor.test.ts b/tests/unit/glm-executor.test.ts
index 4c1a4942352..3d47b375c35 100644
--- a/tests/unit/glm-executor.test.ts
+++ b/tests/unit/glm-executor.test.ts
@@ -164,7 +164,12 @@ test("GlmExecutor separates OpenAI-compatible coding headers from Anthropic head
const anthropicHeaders = executor.buildHeaders(
{
apiKey: "glm-key",
- providerSpecificData: { baseUrl: "https://api.z.ai/api/anthropic/v1/messages" },
+ providerSpecificData: {
+ baseUrl: "https://api.z.ai/api/anthropic/v1/messages",
+ // Same #10798 signature change — Anthropic transport via
+ // providerSpecificData (baseUrl is anthropic-shaped anyway).
+ primaryTransport: "anthropic",
+ },
},
true,
null,
@@ -191,6 +196,8 @@ test("GlmExecutor preserves extra API key rotation", () => {
connectionId: "glm-rotation-test",
providerSpecificData: {
baseUrl: "https://api.z.ai/api/anthropic/v1/messages",
+ // #10798 signature change — Anthropic transport via providerSpecificData.
+ primaryTransport: "anthropic",
extraApiKeys: ["extra-key"],
},
},
@@ -426,10 +433,9 @@ test("GlmExecutor falls back internally to Anthropic transport and returns OpenA
assert.equal(calls[0].url, "https://api.z.ai/api/coding/paas/v4/chat/completions");
assert.equal(calls[0].headers.Authorization, "Bearer glm-key");
assert.equal(calls[1].url, "https://api.z.ai/api/anthropic/v1/messages?beta=true");
- const fallbackKey =
- calls[1].headers["x-api-key"] ||
- String(calls[1].headers.Authorization || "").replace(/^Bearer\s+/i, "");
- assert.equal(fallbackKey, "glm-key");
+ assert.equal(calls[1].headers["x-api-key"], "glm-key");
+ assert.equal(calls[1].headers.Authorization, undefined);
+ assert.equal(calls[1].headers["anthropic-version"], "2023-06-01");
assert.equal(calls[1].body.messages[0].role, "user");
assert.equal(calls[1].body._disableToolPrefix, undefined);
assert.equal(result.targetFormat, "openai");
diff --git a/tests/unit/guide-settings-route.test.ts b/tests/unit/guide-settings-route.test.ts
index 735240dc1e9..7516bd0be5c 100644
--- a/tests/unit/guide-settings-route.test.ts
+++ b/tests/unit/guide-settings-route.test.ts
@@ -198,8 +198,14 @@ test("guide-settings POST preserves existing OpenCode config fields while only u
assert.equal(content.provider.omniroute.options.baseURL, "http://my-omni/v1");
assert.ok(content.provider.omniroute.options.apiKey.startsWith("sk-"));
assert.deepEqual(content.provider.omniroute.models, {
- "cx/gpt-5.6-sol": { name: "GPT-5.6 Sol" },
- "opencode-go/kimi-k2.6": { name: "Kimi K2.6" },
+ "cx/gpt-5.6-sol": {
+ name: "GPT-5.6 Sol",
+ limit: { context: 128_000, output: 8192 },
+ },
+ "opencode-go/kimi-k2.6": {
+ name: "Kimi K2.6",
+ limit: { context: 128_000, output: 8192 },
+ },
});
});
diff --git a/tests/unit/hard-session-lease-bypass-inventory.test.ts b/tests/unit/hard-session-lease-bypass-inventory.test.ts
index bafa2971460..396c54015c7 100644
--- a/tests/unit/hard-session-lease-bypass-inventory.test.ts
+++ b/tests/unit/hard-session-lease-bypass-inventory.test.ts
@@ -86,7 +86,7 @@ const EXPECTED: Record> = {
"src/app/api/providers/client/route.ts": 1,
"src/app/api/providers/free-onboarding/route.ts": 2,
"src/app/api/providers/import/route.ts": 1,
- "src/app/api/providers/route.ts": 4,
+ "src/app/api/providers/route.ts": 2,
"src/app/api/providers/test-batch/route.ts": 2,
"src/app/api/rate-limits/route.ts": 1,
"src/app/api/services/dario/admin/import-from-omniroute/route.ts": 2,
diff --git a/tests/unit/lkgp-enabled-context-11181.test.ts b/tests/unit/lkgp-enabled-context-11181.test.ts
new file mode 100644
index 00000000000..04950c2c2a5
--- /dev/null
+++ b/tests/unit/lkgp-enabled-context-11181.test.ts
@@ -0,0 +1,141 @@
+/**
+ * #11181 — the `lkgpEnabled` settings toggle must actually reach RoutingContext.
+ *
+ * `LKGPStrategyImpl.select()` guards with `context.lkgpEnabled === false`
+ * (open-sse/services/autoCombo/routerStrategy.ts), and the setting is persisted
+ * by the Routing settings tab (src/shared/validation/settingsSchemas.ts). But the
+ * RoutingContext literal built in resolveAutoStrategyOrder() never carried the
+ * field, so the guard could never fire in production.
+ *
+ * tests/unit/router-strategies.test.ts already covers the guard — but it hands
+ * the strategy a context it built itself, so it stays green whether or not the
+ * production construction site populates the field. These tests drive
+ * resolveAutoStrategyOrder() with a *persisted* setting instead, which is the
+ * only level at which the wiring is observable.
+ */
+import { test, after } from "node:test";
+import assert from "node:assert/strict";
+
+import fs from "node:fs";
+import os from "node:os";
+import path from "node:path";
+
+const TEST_DATA_DIR = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-lkgp-11181-"));
+process.env.DATA_DIR = TEST_DATA_DIR;
+
+const { resolveAutoStrategyOrder } =
+ await import("@omniroute/open-sse/services/combo/resolveAutoStrategy.ts");
+const settingsDb = await import("@/lib/db/settings.ts");
+const { resetDbInstance } = await import("@/lib/db/core.ts");
+
+after(() => {
+ resetDbInstance();
+});
+
+const target = (provider: string, modelStr: string): never =>
+ ({
+ kind: "model",
+ stepId: "s1",
+ executionKey: `${provider}>${modelStr}`,
+ modelStr,
+ provider,
+ providerId: null,
+ connectionId: null,
+ weight: 1,
+ label: null,
+ }) as never;
+
+const candidate = (provider: string, model: string, overrides: Record = {}) => ({
+ kind: "model",
+ stepId: "s1",
+ executionKey: `${provider}>${model}`,
+ modelStr: model,
+ provider,
+ model,
+ quotaRemaining: 100,
+ quotaTotal: 100,
+ circuitBreakerState: "CLOSED",
+ costPer1MTokens: 1,
+ p95LatencyMs: 100,
+ latencyStdDev: 10,
+ errorRate: 0,
+ ...overrides,
+});
+
+// "cheap" wins under the rules scorer (cheapest + fastest + most stable);
+// "pricey" only ever wins by being the persisted last-known-good provider.
+const candidates = () =>
+ [
+ candidate("openai", "cheap-model", {
+ costPer1MTokens: 0.01,
+ p95LatencyMs: 10,
+ latencyStdDev: 1,
+ }),
+ candidate("anthropic", "pricey-model", {
+ costPer1MTokens: 50,
+ p95LatencyMs: 5000,
+ latencyStdDev: 900,
+ }),
+ ] as never;
+
+function capturingLog() {
+ const entries: string[] = [];
+ const push = (_tag: unknown, msg: unknown) => entries.push(String(msg));
+ return { entries, info: push, warn: push, error: push, debug: push };
+}
+
+async function runWithSettings(comboName: string, settings: Record | null) {
+ // The LKGP pin resolveAutoStrategyOrder reads is getLKGP(combo.name, combo.id || combo.name).
+ await settingsDb.setLKGP(comboName, comboName, "anthropic");
+
+ const log = capturingLog();
+ const result = await resolveAutoStrategyOrder({
+ orderedTargets: [target("openai", "cheap-model"), target("anthropic", "pricey-model")],
+ body: { messages: [{ role: "user", content: "hi" }] },
+ combo: {
+ id: comboName,
+ name: comboName,
+ autoConfig: {
+ routerStrategy: "lkgp",
+ candidatePool: ["openai", "anthropic"],
+ explorationRate: 0,
+ },
+ },
+ settings,
+ config: {},
+ relayOptions: null,
+ resilienceSettings: { quotaPreflight: { enabled: false } },
+ log,
+ buildAutoCandidates: (async () => candidates()) as never,
+ } as never);
+
+ assert.ok("orderedTargets" in result, "expected an ordering result, not an earlyResponse");
+ const selection = log.entries.find((entry) => entry.startsWith("Auto selection:")) ?? "";
+ return { result, selection };
+}
+
+test("control — with lkgpEnabled unset the LKGP pin still wins (guard must not over-fire)", async () => {
+ const { result, selection } = await runWithSettings("lkgp-11181-default", null);
+ assert.match(selection, /LKGP: using last known good provider anthropic/);
+ if ("orderedTargets" in result) {
+ assert.equal(result.orderedTargets[0].provider, "anthropic");
+ }
+});
+
+test("#11181 — a persisted lkgpEnabled:false makes the lkgp strategy delegate to rules", async () => {
+ const { result, selection } = await runWithSettings("lkgp-11181-disabled", {
+ lkgpEnabled: false,
+ });
+
+ // The whole point of the toggle: the LKGP pin must be ignored and the 6-factor
+ // rules scorer must pick the winner instead.
+ assert.doesNotMatch(
+ selection,
+ /LKGP: using last known good provider/,
+ `lkgpEnabled:false must disable LKGP selection, got: ${selection}`
+ );
+ assert.match(selection, /RulesStrategy: score=/, `expected rules fallback, got: ${selection}`);
+ if ("orderedTargets" in result) {
+ assert.equal(result.orderedTargets[0].provider, "openai");
+ }
+});
diff --git a/tests/unit/logfare-registry.test.ts b/tests/unit/logfare-registry.test.ts
new file mode 100644
index 00000000000..d5439cd1b23
--- /dev/null
+++ b/tests/unit/logfare-registry.test.ts
@@ -0,0 +1,63 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+
+import { logfareProvider } from "../../open-sse/config/providers/registry/logfare/index.ts";
+
+const { APIKEY_PROVIDERS } = await import(
+ "../../src/shared/constants/providers.ts"
+);
+const { REGISTRY: providerRegistry } =
+ await import("../../open-sse/config/providerRegistry.ts");
+const { NAMED_OPENAI_STYLE_PROVIDERS, isNamedOpenAIStyleProvider } =
+ await import(
+ "../../src/app/api/providers/[id]/models/discovery/providerSets.ts"
+ );
+
+const SPEC = {
+ id: "logfare",
+ alias: "logfare",
+ name: "Logfare",
+ website: "https://logfare.ai",
+ chatUrl: "https://logfare.ai/v1/chat/completions",
+ modelsUrl: "https://logfare.ai/v1/models",
+};
+
+test("logfareProvider registry entry has correct configuration", () => {
+ assert.equal(logfareProvider.id, "logfare");
+ assert.equal(logfareProvider.alias, "logfare");
+ assert.equal(logfareProvider.format, "openai");
+ assert.equal(logfareProvider.executor, "default");
+ assert.equal(logfareProvider.baseUrl, SPEC.chatUrl);
+ assert.equal(logfareProvider.modelsUrl, SPEC.modelsUrl);
+ assert.equal(logfareProvider.authType, "apikey");
+ assert.equal(logfareProvider.authHeader, "bearer");
+ // Catalog is discovered live from /v1/models; no hardcoded seed.
+ assert.equal(logfareProvider.passthroughModels, true);
+ assert.equal(logfareProvider.models.length, 0);
+});
+
+test("APIKEY_PROVIDERS.logfare is registered with the canonical identity", () => {
+ const entry = APIKEY_PROVIDERS[SPEC.id];
+ assert.ok(entry, `APIKEY_PROVIDERS.${SPEC.id} must be defined`);
+ assert.equal(entry.id, SPEC.id);
+ assert.equal(entry.alias, SPEC.alias);
+ assert.equal(entry.name, SPEC.name);
+ assert.equal(entry.website, SPEC.website);
+ assert.equal(entry.hasFree, true);
+ assert.equal(typeof entry.freeNote, "string");
+ assert.equal(typeof entry.apiHint, "string");
+ assert.match(entry.color, /^#[0-9A-Fa-f]{6}$/);
+});
+
+test("providerRegistry exposes the OpenAI-compatible chat completions URL", () => {
+ assert.equal(providerRegistry[SPEC.id].baseUrl, SPEC.chatUrl);
+ assert.equal(providerRegistry[SPEC.id].modelsUrl, SPEC.modelsUrl);
+});
+
+test("logfare is classified as a named OpenAI-style provider (live-fetch path)", () => {
+ assert.ok(
+ NAMED_OPENAI_STYLE_PROVIDERS.has(SPEC.id),
+ "logfare must be in NAMED_OPENAI_STYLE_PROVIDERS for live /v1/models fetch"
+ );
+ assert.equal(isNamedOpenAIStyleProvider(SPEC.id), true);
+});
diff --git a/tests/unit/login-11143.test.ts b/tests/unit/login-11143.test.ts
new file mode 100644
index 00000000000..f55f2edecf3
--- /dev/null
+++ b/tests/unit/login-11143.test.ts
@@ -0,0 +1,21 @@
+import assert from "node:assert/strict";
+import { test } from "node:test";
+import fs from "node:fs";
+import path from "node:path";
+
+test("login page performs full window.location navigation after authentication to avoid cookie race", () => {
+ const loginPagePath = path.resolve(process.cwd(), "src/app/login/page.tsx");
+ const content = fs.readFileSync(loginPagePath, "utf8");
+
+ // Ensure router.push("/dashboard") is replaced with window.location.href
+ assert.equal(
+ content.includes('router.push("/dashboard")'),
+ false,
+ "LoginPage should not use router.push('/dashboard') after login"
+ );
+ assert.equal(
+ content.includes('window.location.href = "/dashboard"'),
+ true,
+ "LoginPage must perform full window.location navigation after login"
+ );
+});
diff --git a/tests/unit/memory-system-first-6135.test.ts b/tests/unit/memory-system-first-6135.test.ts
index 7104007ae64..339f83792cf 100644
--- a/tests/unit/memory-system-first-6135.test.ts
+++ b/tests/unit/memory-system-first-6135.test.ts
@@ -49,6 +49,11 @@ describe("injectMemory system-must-be-first (#6135)", () => {
it("flags xiaomi-mimo (and alias mimo) as system-must-be-first", () => {
assert.equal(systemMessageMustBeFirst("xiaomi-mimo"), true);
assert.equal(systemMessageMustBeFirst("mimo"), true);
+ // tokenrouter: confirmed live 2026-08-22 — mid-array system message
+ // (e.g. the purifyHistory compression notice) -> HTTP 400
+ // "System message must be at the beginning".
+ assert.equal(systemMessageMustBeFirst("tokenrouter"), true);
+ assert.equal(systemMessageMustBeFirst("TokenRouter"), true); // case-insensitive
// default: unlisted providers keep current (non-first-constrained) behavior
assert.equal(systemMessageMustBeFirst("anthropic"), false);
assert.equal(systemMessageMustBeFirst(null), false);
diff --git a/tests/unit/mitm-cert-install-mode-9442.test.ts b/tests/unit/mitm-cert-install-mode-9442.test.ts
index 979dee7d4c8..4128719c5cd 100644
--- a/tests/unit/mitm-cert-install-mode-9442.test.ts
+++ b/tests/unit/mitm-cert-install-mode-9442.test.ts
@@ -165,11 +165,14 @@ test("filesystem proof: cp under umask 0077 creates mode 0600 (why the fix is ne
const oldUmask = process.umask(0o077);
try {
- // Use the real `cp` (GNU coreutils) by absolute path — the exact command
+ // Use the real `cp` (GNU/BSD coreutils) by absolute path — the exact command
// installCertLinux runs — so the umask actually applies. Node's
// fs.copyFileSync preserves the source mode, which would mask the bug, and
// the bare `cp` on PATH below is a logging stub from the install tests.
- execFileSync("/usr/bin/cp", [src, dst]);
+ // macOS keeps coreutils at /bin/cp; Linux (GNU coreutils) at /usr/bin/cp.
+ const realCp = ["/usr/bin/cp", "/bin/cp"].find((p) => fs.existsSync(p));
+ assert.ok(realCp, "a real cp binary must exist for this filesystem proof");
+ execFileSync(realCp, [src, dst]);
const mode = fs.statSync(dst).mode & 0o777;
assert.equal(mode, 0o600, "cp under umask 0077 must produce 0600 — the bug this fix repairs");
} finally {
diff --git a/tests/unit/opencode-empty-rejection-rotation.test.ts b/tests/unit/opencode-empty-rejection-rotation.test.ts
new file mode 100644
index 00000000000..784a4b83a74
--- /dev/null
+++ b/tests/unit/opencode-empty-rejection-rotation.test.ts
@@ -0,0 +1,416 @@
+import { describe, it, beforeEach, afterEach, before, after } from "node:test";
+import assert from "node:assert";
+import net from "node:net";
+import { OpencodeExecutor } from "../../open-sse/executors/opencode.ts";
+import type { ExecutorLog, ProviderCredentials } from "../../open-sse/executors/base.ts";
+import { resolveProxyForRequest } from "../../open-sse/utils/proxyFetch.ts";
+import {
+ isEmptyUpstreamRejection,
+ extractChatcmplId,
+} from "../../open-sse/executors/accountRotation.ts";
+
+/**
+ * Empty-upstream-rejection rotation (#design opencode-empty-rejection-rotation).
+ *
+ * An upstream 400 whose body carries no usable completion (the observed malformed
+ * envelope: `choices[0].message` with no error field, no real content,
+ * `finish_reason: null`) must be rotated/retried instead of propagated as a fatal
+ * success — that was killing subagent sessions. These tests pin the wiring:
+ *
+ * 1. A 400 empty rejection rotates to the next account (and its proxy).
+ * 2. The retry budget is bounded: +1 attempt for a single account, exactly N
+ * for an N-account all-empty run (propagate the last 400, never loop forever).
+ * 3. A 400 carrying a real error field (or non-empty content) still propagates
+ * immediately — no cooldown, no success, no rotation.
+ * 4. The 200/success path is never cloned or read (anti-bufferisation).
+ *
+ * The dispatch layer is mocked by stubbing globalThis.fetch (exactly what the
+ * #4954 proxy integration test does). Three throwaway TCP listeners stand in for
+ * the per-account proxies so runWithProxyContext's reachability probe passes.
+ */
+
+const log: ExecutorLog = { debug() {}, info() {}, warn() {}, error() {} };
+
+const ACCOUNT_A = "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa";
+const ACCOUNT_B = "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb";
+const ACCOUNT_C = "cccccccccccccccccccccccccccccccc";
+
+const EMPTY_BODY =
+ '{"id":"chatcmpl_44fn2g6e7kk","object":"chat.completion","created":1787419957,"model":"muse-spark-1.2-contributor-free","choices":[{"index":0,"message":{"role":"assistant"},"finish_reason":null}]}';
+const ERROR_BODY = JSON.stringify({
+ error: { message: "bad request", type: "invalid_request_error" },
+});
+
+let serverA: net.Server;
+let serverB: net.Server;
+let serverC: net.Server;
+let portA = 0;
+let portB = 0;
+let portC = 0;
+
+function listen(server: net.Server): Promise {
+ return new Promise((resolve) => {
+ server.listen(0, "127.0.0.1", () => {
+ resolve((server.address() as net.AddressInfo).port);
+ });
+ });
+}
+
+before(async () => {
+ serverA = net.createServer((s) => s.destroy());
+ serverB = net.createServer((s) => s.destroy());
+ serverC = net.createServer((s) => s.destroy());
+ portA = await listen(serverA);
+ portB = await listen(serverB);
+ portC = await listen(serverC);
+});
+
+after(() => {
+ serverA?.close();
+ serverB?.close();
+ serverC?.close();
+});
+
+function portFor(fp: string): number {
+ if (fp === ACCOUNT_A) return portA;
+ if (fp === ACCOUNT_B) return portB;
+ return portC;
+}
+
+/** `fingerprints` accounts; `proxied` is the subset that get a dedicated proxy
+ * (defaults to all). A proxy-less account shares the default egress. */
+function credentialsFor(
+ fingerprints: string[],
+ proxied: string[] = [...fingerprints]
+): ProviderCredentials {
+ return {
+ apiKey: null,
+ accessToken: null,
+ connectionId: "noauth",
+ providerSpecificData: {
+ fingerprints,
+ ...(proxied.length > 0 && {
+ accountProxies: proxied.map((fp) => ({
+ fingerprint: fp,
+ proxy: { type: "http", host: "127.0.0.1", port: portFor(fp) },
+ })),
+ }),
+ },
+ };
+}
+
+/** A Response subclass that counts clone() so we can assert the executor never
+ * buffers a 200/streaming response. Note: `clone()` returns a plain Response, so
+ * only `clone()` is reliably counted (a read on the clone hits the native
+ * method, not this override) — counting clones is the meaningful invariant. */
+class SpyResponse extends Response {
+ static clones = 0;
+ clone(): Response {
+ SpyResponse.clones++;
+ return super.clone();
+ }
+}
+
+interface PlanStep {
+ status: number;
+ body?: string;
+ throw?: Error;
+}
+
+describe("OpencodeExecutor empty-rejection rotation", () => {
+ let originalFetch: typeof globalThis.fetch;
+ let observed: Array<{ source: string; host: string | null; port: string | null }>;
+ const GUARD_FLAG = "NETWORK_ROTATION_SHARED_EGRESS_GUARD";
+ let savedGuardFlag: string | undefined;
+
+ beforeEach(() => {
+ originalFetch = globalThis.fetch;
+ observed = [];
+ SpyResponse.clones = 0;
+ savedGuardFlag = process.env[GUARD_FLAG];
+ delete process.env[GUARD_FLAG];
+ });
+
+ afterEach(() => {
+ globalThis.fetch = originalFetch;
+ if (savedGuardFlag === undefined) delete process.env[GUARD_FLAG];
+ else process.env[GUARD_FLAG] = savedGuardFlag;
+ });
+
+ function installFetch(plan: PlanStep[]) {
+ let call = 0;
+ globalThis.fetch = (async (input: RequestInfo | URL) => {
+ const url =
+ typeof input === "string" ? input : input instanceof URL ? input.toString() : input.url;
+ const resolved = resolveProxyForRequest(url);
+ observed.push({
+ source: resolved.source,
+ host: resolved.proxyUrl ? new URL(resolved.proxyUrl).hostname : null,
+ port: resolved.proxyUrl ? new URL(resolved.proxyUrl).port : null,
+ });
+ const step = plan[Math.min(call, plan.length - 1)];
+ call++;
+ if (step.throw) throw step.throw;
+ return new SpyResponse(step.body ?? JSON.stringify({ ok: step.status === 200 }), {
+ status: step.status,
+ headers: { "Content-Type": "application/json" },
+ });
+ }) as typeof globalThis.fetch;
+ }
+
+ /**
+ * Launches the executor. Asserts the predicate itself behaves (regression guard
+ * for the design's signature — the wiring tests below depend on it).
+ */
+ it("predicate matches the observed envelope and rejects real errors", () => {
+ assert.strictEqual(isEmptyUpstreamRejection(400, EMPTY_BODY), true);
+ assert.strictEqual(isEmptyUpstreamRejection(200, EMPTY_BODY), false);
+ assert.strictEqual(isEmptyUpstreamRejection(400, ERROR_BODY), false);
+ assert.strictEqual(extractChatcmplId(EMPTY_BODY), "chatcmpl_44fn2g6e7kk");
+ });
+
+ it("rotates to the next account on an empty 400 rejection (loop)", async () => {
+ const exec = new OpencodeExecutor("opencode-zen");
+ installFetch([{ status: 400, body: EMPTY_BODY }, { status: 200 }]);
+
+ const result = await exec.execute({
+ model: "deepseek-v4-flash-free",
+ body: { messages: [{ role: "user", content: "hi" }], stream: false },
+ stream: false,
+ signal: null,
+ credentials: credentialsFor([ACCOUNT_A, ACCOUNT_B]),
+ log,
+ });
+
+ assert.strictEqual(
+ (result as { response: Response }).response.status,
+ 200,
+ "must rotate past the empty 400"
+ );
+ assert.ok(observed.length >= 2, "should have dispatched on a second account");
+ assert.ok(
+ observed.some((o) => o.port === String(portA)),
+ "first attempt on account A"
+ );
+ assert.ok(
+ observed.some((o) => o.port === String(portB)),
+ "rotated attempt on account B"
+ );
+ });
+
+ it("caps an all-empty N-account run at N attempts and propagates the last 400 intact", async () => {
+ const exec = new OpencodeExecutor("opencode-zen");
+ installFetch([
+ { status: 400, body: EMPTY_BODY },
+ { status: 400, body: EMPTY_BODY },
+ { status: 400, body: EMPTY_BODY },
+ { status: 200 },
+ ]);
+
+ const result = await exec.execute({
+ model: "deepseek-v4-flash-free",
+ body: { messages: [{ role: "user", content: "hi" }], stream: false },
+ stream: false,
+ signal: null,
+ credentials: credentialsFor([ACCOUNT_A, ACCOUNT_B, ACCOUNT_C]),
+ log,
+ });
+
+ assert.strictEqual(
+ (result as { response: Response }).response.status,
+ 400,
+ "must propagate the last empty 400"
+ );
+ assert.strictEqual(observed.length, 3, "must NOT exceed N attempts (no infinite loop)");
+ assert.ok(SpyResponse.clones >= 1, "the empty 400 path must read the body to classify it");
+ const propagated = await (result as { response: Response }).response.clone().text();
+ assert.strictEqual(propagated, EMPTY_BODY, "propagated 400 body must stay intact");
+ for (const p of observed) {
+ assert.strictEqual(p.source, "context", "every dispatch must egress through a proxy context");
+ }
+ });
+
+ it("retries the same proxied account once when it is the only account", async () => {
+ const exec = new OpencodeExecutor("opencode-zen");
+ installFetch([
+ { status: 400, body: EMPTY_BODY },
+ { status: 400, body: EMPTY_BODY },
+ ]);
+
+ const result = await exec.execute({
+ model: "deepseek-v4-flash-free",
+ body: { messages: [{ role: "user", content: "hi" }], stream: false },
+ stream: false,
+ signal: null,
+ credentials: credentialsFor([ACCOUNT_A]),
+ log,
+ });
+
+ assert.strictEqual((result as { response: Response }).response.status, 400);
+ assert.strictEqual(observed.length, 2, "exactly one bounded retry on the sole account");
+ assert.ok(
+ observed.every((o) => o.port === String(portA)),
+ "both attempts egress through the single account's proxy"
+ );
+ });
+
+ it("coexists with 429 rotation and 200 success in the same request", async () => {
+ const exec = new OpencodeExecutor("opencode-zen");
+ installFetch([{ status: 429 }, { status: 400, body: EMPTY_BODY }, { status: 200 }]);
+
+ const result = await exec.execute({
+ model: "deepseek-v4-flash-free",
+ body: { messages: [{ role: "user", content: "hi" }], stream: false },
+ stream: false,
+ signal: null,
+ credentials: credentialsFor([ACCOUNT_A, ACCOUNT_B, ACCOUNT_C]),
+ log,
+ });
+
+ assert.strictEqual(
+ (result as { response: Response }).response.status,
+ 200,
+ "final response should succeed"
+ );
+ assert.strictEqual(observed.length, 3, "429 + empty-400 + success across three accounts");
+ assert.ok(
+ observed.some((o) => o.port === String(portA)),
+ "account A (429)"
+ );
+ assert.ok(
+ observed.some((o) => o.port === String(portB)),
+ "account B (empty 400)"
+ );
+ assert.ok(
+ observed.some((o) => o.port === String(portC)),
+ "account C (200)"
+ );
+ });
+
+ it("propagates a 400 carrying an error field immediately (no rotation)", async () => {
+ const exec = new OpencodeExecutor("opencode-zen");
+ installFetch([{ status: 400, body: ERROR_BODY }]);
+
+ const result = await exec.execute({
+ model: "deepseek-v4-flash-free",
+ body: { messages: [{ role: "user", content: "hi" }], stream: false },
+ stream: false,
+ signal: null,
+ credentials: credentialsFor([ACCOUNT_A, ACCOUNT_B]),
+ log,
+ });
+
+ assert.strictEqual(
+ (result as { response: Response }).response.status,
+ 400,
+ "real error 400 must propagate"
+ );
+ assert.strictEqual(observed.length, 1, "must NOT rotate on a genuine error 400");
+ });
+
+ it("never clones or reads the body of a 200 via the loop", async () => {
+ const exec = new OpencodeExecutor("opencode-zen");
+ installFetch([{ status: 200 }, { status: 200 }, { status: 200 }]);
+
+ const result = await exec.execute({
+ model: "deepseek-v4-flash-free",
+ body: { messages: [{ role: "user", content: "hi" }], stream: false },
+ stream: false,
+ signal: null,
+ credentials: credentialsFor([ACCOUNT_A, ACCOUNT_B, ACCOUNT_C]),
+ log,
+ });
+
+ assert.strictEqual((result as { response: Response }).response.status, 200);
+ assert.strictEqual(SpyResponse.clones, 0, "loop 200 must never be cloned");
+ });
+
+ it("retries once via the fast path when a direct account answers an empty 400", async () => {
+ const exec = new OpencodeExecutor("opencode-zen");
+ installFetch([
+ { status: 400, body: EMPTY_BODY },
+ { status: 400, body: EMPTY_BODY },
+ ]);
+
+ const result = await exec.execute({
+ model: "deepseek-v4-flash-free",
+ body: { messages: [{ role: "user", content: "hi" }], stream: false },
+ stream: false,
+ signal: null,
+ credentials: credentialsFor([ACCOUNT_A], []),
+ log,
+ });
+
+ assert.strictEqual((result as { response: Response }).response.status, 400);
+ assert.strictEqual(observed.length, 2, "fast path must retry the direct account exactly once");
+ });
+
+ it("propagates the second 400 intact when the fast path retries and empty-rejects again", async () => {
+ const exec = new OpencodeExecutor("opencode-zen");
+ installFetch([
+ { status: 400, body: EMPTY_BODY },
+ { status: 400, body: EMPTY_BODY },
+ ]);
+
+ const result = await exec.execute({
+ model: "deepseek-v4-flash-free",
+ body: { messages: [{ role: "user", content: "hi" }], stream: false },
+ stream: false,
+ signal: null,
+ credentials: credentialsFor([ACCOUNT_A], []),
+ log,
+ });
+
+ assert.strictEqual((result as { response: Response }).response.status, 400);
+ const propagated = await (result as { response: Response }).response.clone().text();
+ assert.strictEqual(propagated, EMPTY_BODY, "second rejection propagates with intact body");
+ assert.strictEqual(observed.length, 2, "exactly one retry, no loop");
+ });
+
+ it("never clones or reads the body of a 200 via the fast path", async () => {
+ const exec = new OpencodeExecutor("opencode-zen");
+ installFetch([{ status: 200 }]);
+
+ const result = await exec.execute({
+ model: "deepseek-v4-flash-free",
+ body: { messages: [{ role: "user", content: "hi" }], stream: false },
+ stream: false,
+ signal: null,
+ credentials: credentialsFor([ACCOUNT_A], []),
+ log,
+ });
+
+ assert.strictEqual((result as { response: Response }).response.status, 200);
+ assert.strictEqual(SpyResponse.clones, 0, "fast path 200 must never be cloned");
+ });
+
+ it("rotates to a proxied account after a proxy-less account empty-rejects (shared-egress guard on by default)", async () => {
+ const exec = new OpencodeExecutor("opencode-zen");
+ installFetch([{ status: 400, body: EMPTY_BODY }, { status: 200 }]);
+
+ const result = await exec.execute({
+ model: "deepseek-v4-flash-free",
+ body: { messages: [{ role: "user", content: "hi" }], stream: false },
+ stream: false,
+ signal: null,
+ // A proxy-less, B proxied: B must still be tried and succeed.
+ credentials: credentialsFor([ACCOUNT_A, ACCOUNT_B], [ACCOUNT_B]),
+ log,
+ });
+
+ assert.strictEqual(
+ (result as { response: { status: number } }).response.status,
+ 200,
+ "the proxied account (B) must still be tried and must succeed"
+ );
+ assert.strictEqual(observed.length, 2, "exactly one empty rejection (A) then one success (B)");
+ assert.ok(
+ observed.some((o) => o.source === "direct"),
+ "first dispatch on the proxy-less account"
+ );
+ assert.ok(
+ observed.some((o) => o.port === String(portB)),
+ "rotated dispatch on the proxied account"
+ );
+ });
+});
diff --git a/tests/unit/openrouter-free-model-credits-exhausted.test.ts b/tests/unit/openrouter-free-model-credits-exhausted.test.ts
index 601553b658b..129d2fac6d9 100644
--- a/tests/unit/openrouter-free-model-credits-exhausted.test.ts
+++ b/tests/unit/openrouter-free-model-credits-exhausted.test.ts
@@ -78,7 +78,11 @@ test("getProviderCredentials still refuses a PAID OpenRouter model on a credits_
"anthropic/claude-opus-4.5"
);
- assert.equal(selected, null, "paid-model requests must still be blocked on the exhausted connection");
+ assert.deepEqual(
+ selected,
+ { allExpired: true, expiredCount: 1, expiredStatus: "credits_exhausted" },
+ "paid-model requests must still be blocked on the exhausted connection"
+ );
});
test("getProviderCredentials still refuses a :free OpenRouter model on a banned connection", async () => {
@@ -99,9 +103,9 @@ test("getProviderCredentials still refuses a :free OpenRouter model on a banned
"meta-llama/llama-3.1-8b-instruct:free"
);
- assert.equal(
+ assert.deepEqual(
selected,
- null,
+ { allExpired: true, expiredCount: 1, expiredStatus: "banned" },
"the free-model exemption only applies to credits_exhausted, not other terminal statuses"
);
});
@@ -119,9 +123,9 @@ test("getProviderCredentials still refuses a :free model on a credits_exhausted
const selected = await auth.getProviderCredentials("openai", null, null, "some-model:free");
- assert.equal(
+ assert.deepEqual(
selected,
- null,
+ { allExpired: true, expiredCount: 1, expiredStatus: "credits_exhausted" },
"the exemption is OpenRouter-specific, since only OpenRouter uses the :free naming convention with a shared balance"
);
});
diff --git a/tests/unit/perf-a-b-c-d.test.ts b/tests/unit/perf-a-b-c-d.test.ts
new file mode 100644
index 00000000000..0a131e0a34d
--- /dev/null
+++ b/tests/unit/perf-a-b-c-d.test.ts
@@ -0,0 +1,26 @@
+import assert from "node:assert/strict";
+import { test } from "node:test";
+import { logProxyEvent, flushProxyLogsSync } from "../../src/lib/proxyLogger.ts";
+import { shouldStripCloudCodeThinking } from "../../open-sse/services/cloudCodeThinking.ts";
+
+test("Part A: async proxy log batching queues entries without synchronous failure", () => {
+ const sampleLog = {
+ status: "success",
+ provider: "test-provider",
+ latencyMs: 15,
+ };
+
+ const logged = logProxyEvent(sampleLog);
+ assert.equal(logged.provider, "test-provider");
+ assert.equal(typeof logged.id, "string");
+
+ // Ensure flush completes without throwing
+ assert.doesNotThrow(() => {
+ flushProxyLogsSync();
+ });
+});
+
+test("Part D: pre-compiled regex in cloudCodeThinking model normalization", () => {
+ assert.equal(shouldStripCloudCodeThinking("antigravity", "antigravity/claude-3-7-sonnet"), true);
+ assert.equal(shouldStripCloudCodeThinking("antigravity", "models/gemini-2.5-pro"), false);
+});
diff --git a/tests/unit/provider-sweep-live-discovery.test.ts b/tests/unit/provider-sweep-live-discovery.test.ts
index 323a1599f3e..2431179fa7a 100644
--- a/tests/unit/provider-sweep-live-discovery.test.ts
+++ b/tests/unit/provider-sweep-live-discovery.test.ts
@@ -48,7 +48,7 @@ interface ModelsBody {
}
// provider → the upstream /models URL the route resolves from its registry baseUrl.
-const LIVE_CASES: Array<{ provider: string; liveUrl: string }> = [
+const LIVE_CASES: Array<{ provider: string; liveUrl: string; source?: string }> = [
{ provider: "venice", liveUrl: "https://api.venice.ai/api/v1/models" },
{ provider: "deepinfra", liveUrl: "https://api.deepinfra.com/v1/openai/models" },
{ provider: "wandb", liveUrl: "https://api.inference.wandb.ai/v1/models" },
@@ -62,7 +62,11 @@ const LIVE_CASES: Array<{ provider: string; liveUrl: string }> = [
{ provider: "ovhcloud", liveUrl: "https://oai.endpoints.kepler.ai.cloud.ovh.net/v1/models" },
{ provider: "sambanova", liveUrl: "https://api.sambanova.ai/v1/models" },
{ provider: "orcarouter", liveUrl: "https://api.orcarouter.ai/v1/models" },
- { provider: "uncloseai", liveUrl: "https://hermes.ai.unturf.com/v1/models" },
+ {
+ provider: "uncloseai",
+ liveUrl: "https://hermes.ai.unturf.com/v1/models",
+ source: "upstream",
+ },
{ provider: "opencode-go", liveUrl: "https://opencode.ai/zen/go/v1/models" },
{ provider: "baseten", liveUrl: "https://inference.baseten.co/v1/models" },
{ provider: "hyperbolic", liveUrl: "https://api.hyperbolic.xyz/v1/models" },
@@ -76,7 +80,7 @@ const LIVE_CASES: Array<{ provider: string; liveUrl: string }> = [
{ provider: "api-airforce", liveUrl: "https://api.airforce/v1/models" },
];
-for (const { provider, liveUrl } of LIVE_CASES) {
+for (const { provider, liveUrl, source = "api" } of LIVE_CASES) {
test(`sweep: ${provider} import fetches the live /models catalog`, async () => {
await resetStorage();
const connection = await providersDb.createProviderConnection({
@@ -109,7 +113,7 @@ for (const { provider, liveUrl } of LIVE_CASES) {
const body = (await response.json()) as ModelsBody;
assert.equal(body.provider, provider);
assert.ok(fetched, `should have probed ${liveUrl}`);
- assert.equal(body.source, "api", "should serve the live upstream catalog, not local_catalog");
+ assert.equal(body.source, source, "should serve the live upstream catalog, not local_catalog");
const ids = body.models.map((m) => m.id);
assert.ok(
ids.includes(`${provider}-live-a`) && ids.includes(`${provider}-live-b`),
diff --git a/tests/unit/providers-constants-split.test.ts b/tests/unit/providers-constants-split.test.ts
index 3eabac24bb9..e22b3fd4e83 100644
--- a/tests/unit/providers-constants-split.test.ts
+++ b/tests/unit/providers-constants-split.test.ts
@@ -25,7 +25,7 @@
// #10729) brings it to 229; Token Kiosk (gateways, #10722) — merged in the same
// merge-train batch — independently bumped the gateways family too, landing at 231; Freebuff
// (gateways, #10531) brings it to 232. #8864 moves uncloseai (gateways family) into
-// NOAUTH_PROVIDERS, dropping the APIKEY_PROVIDERS count to 231.
+// NOAUTH_PROVIDERS, dropping the APIKEY_PROVIDERS count to 231. Logfare (gateways, #10987) brings it back to 232.
import { test } from "node:test";
import assert from "node:assert/strict";
@@ -54,12 +54,12 @@ test("barrel still exports every catalog + key helpers", () => {
}
});
-test("APIKEY_PROVIDERS merges the 6 family files into 231 entries (no loss / no dup)", async () => {
+test("APIKEY_PROVIDERS merges the 6 family files into 232 entries (no loss / no dup)", async () => {
const keys = Object.keys((P as Record).APIKEY_PROVIDERS);
- assert.equal(keys.length, 231);
- assert.equal(new Set(keys).size, 231, "duplicate keys after spread-merge");
+ assert.equal(keys.length, 232);
+ assert.equal(new Set(keys).size, 232, "duplicate keys after spread-merge");
// the merged object's entry-count equals the sum of the 6 semantic family files; families are a
- // strict partition (every provider in exactly one), so the sum must be exactly 231.
+ // strict partition (every provider in exactly one), so the sum must be exactly 232.
const families: [string, string][] = [
["gateways", "APIKEY_PROVIDERS_GATEWAYS"],
["frontier-labs", "APIKEY_PROVIDERS_FRONTIER"],
@@ -79,7 +79,7 @@ test("APIKEY_PROVIDERS merges the 6 family files into 231 entries (no loss / no
seen.add(k);
}
}
- assert.equal(famTotal, 231, "families must partition all 231 providers");
+ assert.equal(famTotal, 232, "families must partition all 232 providers");
});
test("AI_PROVIDERS Proxy aggregates all sections; lookups resolve", () => {
diff --git a/tests/unit/proxy-fetch-dns-retry-10443.test.ts b/tests/unit/proxy-fetch-dns-retry-10443.test.ts
new file mode 100644
index 00000000000..2518f58141a
--- /dev/null
+++ b/tests/unit/proxy-fetch-dns-retry-10443.test.ts
@@ -0,0 +1,27 @@
+import assert from "node:assert/strict";
+import { test } from "node:test";
+
+test("proxyFetch identifies transient DNS and network errors (EAI_AGAIN, ENOTFOUND, ECONNREFUSED) as retryable dispatcher errors", () => {
+ const isRetryableError = (err: unknown): boolean => {
+ const msg = err instanceof Error ? err.message : String(err);
+ const errCode = (err as { code?: unknown })?.code;
+ return Boolean(
+ msg.includes("fetch failed") ||
+ errCode === "ECONNREFUSED" ||
+ msg.includes("ECONNREFUSED") ||
+ errCode === "EAI_AGAIN" ||
+ msg.includes("EAI_AGAIN") ||
+ errCode === "ENOTFOUND" ||
+ msg.includes("ENOTFOUND") ||
+ errCode === "ETIMEDOUT" ||
+ msg.includes("ETIMEDOUT") ||
+ (typeof errCode === "string" && errCode.startsWith("UND_ERR")) ||
+ msg.includes("UND_ERR")
+ );
+ };
+
+ assert.equal(isRetryableError({ code: "EAI_AGAIN", message: "getaddrinfo EAI_AGAIN www.googleapis.com" }), true);
+ assert.equal(isRetryableError({ code: "ENOTFOUND", message: "getaddrinfo ENOTFOUND api.example.com" }), true);
+ assert.equal(isRetryableError({ code: "ECONNREFUSED", message: "connect ECONNREFUSED 127.0.0.1:20128" }), true);
+ assert.equal(isRetryableError(new Error("HTTP 404 Not Found")), false);
+});
diff --git a/tests/unit/rateLimitManager-mintime-floor-9763.test.ts b/tests/unit/rateLimitManager-mintime-floor-9763.test.ts
new file mode 100644
index 00000000000..2c4ee557552
--- /dev/null
+++ b/tests/unit/rateLimitManager-mintime-floor-9763.test.ts
@@ -0,0 +1,88 @@
+import test from "node:test";
+import assert from "node:assert/strict";
+
+const rlm = await import("../../open-sse/services/rateLimitManager.ts");
+const {
+ enableRateLimitProtection,
+ withRateLimit,
+ updateFromHeaders,
+ applyRequestQueueSettings,
+ __setLimiterFactoryForTests,
+ __resetRateLimitManagerForTests,
+} = rlm;
+
+test.beforeEach(async () => {
+ await __resetRateLimitManagerForTests();
+});
+
+test("headroom relaxation respects operator minTimeBetweenRequestsMs floor (#9763)", async () => {
+ // Apply an operator-configured minTime floor of 200ms
+ await applyRequestQueueSettings({
+ minTimeBetweenRequestsMs: 200,
+ concurrentRequests: 0,
+ requestsPerMinute: 0,
+ maxWaitMs: 30000,
+ autoEnableApiKeyProvider: false,
+ });
+
+ let capturedMinTime: number | undefined;
+
+ // Inject a fake limiter whose updateSettings captures the minTime.
+ // eslint-disable-next-line @typescript-eslint/no-explicit-any
+ const noop = (): any => undefined;
+
+ __setLimiterFactoryForTests(() => {
+ const listeners: Record void>> = {};
+ const fake = {
+ // eslint-disable-next-line @typescript-eslint/no-explicit-any
+ updateSettings(updates: Record) {
+ capturedMinTime = typeof updates.minTime === "number" ? updates.minTime : undefined;
+ return fake;
+ },
+ on(event: string, fn: (...args: unknown[]) => void) {
+ (listeners[event] ??= []).push(fn);
+ return fake;
+ },
+ schedule(arg0: unknown, arg1?: unknown) {
+ const fn = typeof arg1 === "function" ? arg1 : typeof arg0 === "function" ? arg0 : noop;
+ return fn();
+ },
+ disconnect() {
+ return Promise.resolve();
+ },
+ chain() {
+ return fake;
+ },
+ counts() {
+ return { RECEIVED: 0, QUEUED: 0, RUNNING: 0, EXECUTING: 0 };
+ },
+ currentReservoir() {
+ return Promise.resolve(null);
+ },
+ stop() {
+ return Promise.resolve();
+ },
+ };
+ return fake;
+ });
+
+ enableRateLimitProtection("test-mintime-floor");
+
+ // Materialize the limiter with a dummy request
+ await withRateLimit("openai", "test-mintime-floor", "gpt-4", async () => "ok");
+
+ // Simulate a response with plenty of headroom: remaining=80 > limit*0.5=50
+ const headers = new Headers({
+ "x-ratelimit-limit-requests": "100",
+ "x-ratelimit-remaining-requests": "80",
+ });
+ updateFromHeaders("openai", "test-mintime-floor", headers, 200, "gpt-4");
+
+ // The operator configured minTime=200, so headroom relaxation MUST NOT
+ // override it to 0. Before the fix, capturedMinTime === 0 (RED).
+ assert.strictEqual(
+ capturedMinTime,
+ 200,
+ `Expected minTime=200 (operator floor), got ${capturedMinTime}`
+ );
+});
diff --git a/tests/unit/readyz-route.test.ts b/tests/unit/readyz-route.test.ts
index ec9ed1cb12b..ef5c0f67ee5 100644
--- a/tests/unit/readyz-route.test.ts
+++ b/tests/unit/readyz-route.test.ts
@@ -49,9 +49,11 @@ test("/readyz is omitted from the centralized auth proxy matcher", () => {
assert.equal(/["']\/healthz/.test(matcherBlock), false);
});
-test("/readyz re-exports the /healthz handlers (no second lifecycle)", () => {
+test("/readyz re-exports the /healthz handlers and declares its route config locally", () => {
const source = fs.readFileSync("src/app/readyz/route.ts", "utf8");
assert.match(source, /from ["']\.\.\/healthz\/route["']/);
+ assert.match(source, /export const dynamic = ["']force-dynamic["']/);
+ assert.doesNotMatch(source, /export\s*\{[^}]*\bdynamic\b[^}]*\}\s*from/);
assert.equal(/monitoring/i.test(source), false);
assert.equal(/sqlite/i.test(source), false);
});
diff --git a/tests/unit/reasoning-cache.test.ts b/tests/unit/reasoning-cache.test.ts
index aded469ca85..db538a80314 100644
--- a/tests/unit/reasoning-cache.test.ts
+++ b/tests/unit/reasoning-cache.test.ts
@@ -752,47 +752,70 @@ describe("Reasoning Replay Cache — Translator Replay", () => {
assert.equal(lookupReasoning(callId), "Authentic provider reasoning");
});
- it("should never cache Responses summaries or opaque plaintext companions", () => {
- for (const [suffix, reasoningItem] of [
- [
- "summary",
- {
- type: "reasoning",
- summary: [{ type: "summary_text", text: "Display-only summary" }],
- },
- ],
- [
- "mixed",
- {
- type: "reasoning",
- encrypted_content: "opaque-provider-state",
- content: [{ type: "reasoning_text", text: "Unsafe plaintext companion" }],
- summary: [{ type: "summary_text", text: "Display-only mixed summary" }],
- },
- ],
- ] as const) {
- clearReasoningCacheAll();
- const callId = `call_nonstream_${suffix}_reasoning`;
- const translated = translateNonStreamingResponse(
- {
- object: "response",
- model: "deepseek-v4-flash",
- output: [
- reasoningItem,
- { type: "function_call", call_id: callId, name: "read_file", arguments: "{}" },
- ],
- },
- FORMATS.OPENAI_RESPONSES,
- FORMATS.OPENAI
- ) as { choices?: Array<{ message?: Record }> };
- const message = translated.choices?.[0]?.message;
-
- assert.ok(message);
- assert.equal(message.reasoning_content, undefined);
- assert.ok(Array.isArray(message.reasoning_summary));
- assert.equal(cacheReasoningFromAssistantMessage(message, "deepseek", "deepseek-v4-flash"), 0);
- assert.equal(lookupReasoning(callId), null);
- }
+ it("preserves plaintext reasoning from a mixed plaintext + encrypted_content item (#10949)", () => {
+ clearReasoningCacheAll();
+ const callId = "call_nonstream_mixed_reasoning";
+ const translated = translateNonStreamingResponse(
+ {
+ object: "response",
+ model: "deepseek-v4-flash",
+ output: [
+ {
+ type: "reasoning",
+ content: [
+ {
+ type: "reasoning_text",
+ text: "Let me start by reading the directory to understand the structure of the corpus.",
+ },
+ ],
+ encrypted_content: "",
+ summary: [],
+ },
+ { type: "function_call", call_id: callId, name: "read_file", arguments: "{}" },
+ ],
+ },
+ FORMATS.OPENAI_RESPONSES,
+ FORMATS.OPENAI
+ ) as { choices?: Array<{ message?: Record }> };
+ const message = translated.choices?.[0]?.message;
+
+ assert.ok(message);
+ assert.equal(
+ message.reasoning_content,
+ "Let me start by reading the directory to understand the structure of the corpus."
+ );
+ assert.equal(cacheReasoningFromAssistantMessage(message, "deepseek", "deepseek-v4-flash"), 1);
+ assert.equal(
+ lookupReasoning(callId),
+ "Let me start by reading the directory to understand the structure of the corpus."
+ );
+ });
+
+ it("should never cache summary-only Responses reasoning", () => {
+ clearReasoningCacheAll();
+ const callId = "call_nonstream_summary_reasoning";
+ const translated = translateNonStreamingResponse(
+ {
+ object: "response",
+ model: "deepseek-v4-flash",
+ output: [
+ {
+ type: "reasoning",
+ summary: [{ type: "summary_text", text: "Display-only summary" }],
+ },
+ { type: "function_call", call_id: callId, name: "read_file", arguments: "{}" },
+ ],
+ },
+ FORMATS.OPENAI_RESPONSES,
+ FORMATS.OPENAI
+ ) as { choices?: Array<{ message?: Record }> };
+ const message = translated.choices?.[0]?.message;
+
+ assert.ok(message);
+ assert.equal(message.reasoning_content, undefined);
+ assert.ok(Array.isArray(message.reasoning_summary));
+ assert.equal(cacheReasoningFromAssistantMessage(message, "deepseek", "deepseek-v4-flash"), 0);
+ assert.equal(lookupReasoning(callId), null);
});
it("should preserve client-provided reasoning content", () => {
diff --git a/tests/unit/rejected-request-usage.test.ts b/tests/unit/rejected-request-usage.test.ts
index a25ada202b2..300e77e013c 100644
--- a/tests/unit/rejected-request-usage.test.ts
+++ b/tests/unit/rejected-request-usage.test.ts
@@ -60,12 +60,17 @@ test("gate-rejected request is attributed to the api key in usage_history", asyn
assert.equal(keyRows.length, 1, "expected one usage_history row for the rejected request");
assert.equal(keyRows[0].success, false, "rejected request must be recorded as success:false");
- // call_logs visibility is preserved (dashboard/logs).
- const logs = await callLogs.getCallLogs({});
- const rejected = (logs.logs ?? logs).filter?.(
- (l: { apiKeyName?: string | null }) => l.apiKeyName === "opencode-mac"
- );
- assert.ok(rejected && rejected.length >= 1, "expected a call_logs row for the rejected request");
+ // call_logs visibility is preserved (dashboard/logs). saveCallLog is
+ // fire-and-forget inside recordRejectedRequestUsage, so poll briefly for the
+ // row instead of asserting synchronously after the await.
+ let rejected: Array<{ apiKeyName?: string | null }> = [];
+ for (let i = 0; i < 50 && rejected.length === 0; i++) {
+ const logs = await callLogs.getCallLogs({});
+ const list = (logs.logs ?? logs) as Array<{ apiKeyName?: string | null }>;
+ rejected = (list ?? []).filter((l) => l.apiKeyName === "opencode-mac");
+ if (rejected.length === 0) await new Promise((r) => setTimeout(r, 10));
+ }
+ assert.ok(rejected.length >= 1, "expected a call_logs row for the rejected request");
});
test("combo-exhausted rejection is also counted per api key", async () => {
@@ -111,10 +116,15 @@ test("combo-exhausted rejection persists the client request body for dashboard i
requestBody: { model: "default", messages: [{ role: "user", content: "hello" }] },
});
- const logs = await callLogs.getCallLogs({});
- const rejected = (logs.logs ?? logs).find?.(
- (l: { apiKeyName?: string | null }) => l.apiKeyName === "request-body-test"
- );
+ // saveCallLog is fire-and-forget — poll briefly for the row.
+ let rejected: { id: string; hasRequestBody: boolean } | undefined;
+ for (let i = 0; i < 50 && !rejected; i++) {
+ const logs = await callLogs.getCallLogs({});
+ const list = (logs.logs ?? logs) as Array<{ apiKeyName?: string | null }>;
+ const found = (list ?? []).find((l) => l.apiKeyName === "request-body-test");
+ if (found) rejected = found as unknown as { id: string; hasRequestBody: boolean };
+ else await new Promise((r) => setTimeout(r, 10));
+ }
assert.ok(rejected, "expected a call_logs row for the rejected request");
assert.equal(rejected.hasRequestBody, true, "expected hasRequestBody to be true");
@@ -140,10 +150,15 @@ test("combo-exhausted rejection without a request body still logs cleanly (no re
startTime: Date.now() - 100,
});
- const logs = await callLogs.getCallLogs({});
- const rejected = (logs.logs ?? logs).find?.(
- (l: { apiKeyName?: string | null }) => l.apiKeyName === "no-body-test"
- );
+ // saveCallLog is fire-and-forget — poll briefly for the row.
+ let rejected: { id: string } | undefined;
+ for (let i = 0; i < 50 && !rejected; i++) {
+ const logs = await callLogs.getCallLogs({});
+ const list = (logs.logs ?? logs) as Array<{ apiKeyName?: string | null }>;
+ const found = (list ?? []).find((l) => l.apiKeyName === "no-body-test");
+ if (found) rejected = found as unknown as { id: string };
+ else await new Promise((r) => setTimeout(r, 10));
+ }
assert.ok(rejected, "expected a call_logs row even without a request body");
assert.equal(rejected.hasRequestBody, false);
});
diff --git a/tests/unit/security-route-guard-tiers.test.ts b/tests/unit/security-route-guard-tiers.test.ts
new file mode 100644
index 00000000000..dfc1f3c026f
--- /dev/null
+++ b/tests/unit/security-route-guard-tiers.test.ts
@@ -0,0 +1,10 @@
+import assert from "node:assert/strict";
+import { test } from "node:test";
+import { isLocalOnlyPath } from "../../src/server/authz/routeGuard.ts";
+
+test("isLocalOnlyPath correctly classifies process-spawning endpoints under Tier 1 LOCAL_ONLY", () => {
+ assert.equal(isLocalOnlyPath("/api/services/dario/start"), true);
+ assert.equal(isLocalOnlyPath("/api/mcp/stream"), true);
+ assert.equal(isLocalOnlyPath("/api/cli-tools/runtime/status"), true);
+ assert.equal(isLocalOnlyPath("/api/v1/chat/completions"), false);
+});
diff --git a/tests/unit/services/ServiceSupervisor.test.ts b/tests/unit/services/ServiceSupervisor.test.ts
index 011395733ad..6e04455e614 100644
--- a/tests/unit/services/ServiceSupervisor.test.ts
+++ b/tests/unit/services/ServiceSupervisor.test.ts
@@ -18,6 +18,9 @@ const TEST_DATA_DIR = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-superviso
process.env.DATA_DIR = TEST_DATA_DIR;
process.env.NODE_ENV = "test";
process.env.DISABLE_SQLITE_AUTO_BACKUP = "true";
+// Adoption is intentionally opt-in after GHSA-wg9p-6m2g-4v27. These tests
+// exercise the explicit adoption path, so enable it for this isolated process.
+process.env.OMNIROUTE_ADOPT_EXISTING_SERVICE = "1";
// Import DB core first to trigger migration (creates version_manager with new columns)
const core = await import("../../../src/lib/db/core.ts");
diff --git a/tests/unit/services/portProbePid.test.ts b/tests/unit/services/portProbePid.test.ts
index b88383fd801..c58bf86045c 100644
--- a/tests/unit/services/portProbePid.test.ts
+++ b/tests/unit/services/portProbePid.test.ts
@@ -64,6 +64,12 @@ test("parseNetstatPid matches on the local address, not the foreign one", () =>
assert.equal(parseNetstatPid(stdout, 20128), 596922);
});
+test("parseNetstatPid reads macOS process:pid output", () => {
+ const stdout =
+ "tcp4 0 0 127.0.0.1.20128 *.* LISTEN 0 0 131072 131072 node:596922 00100\n";
+ assert.equal(parseNetstatPid(stdout, 20128), 596922);
+});
+
test("parseNetstatPid ignores non-listening rows and unknown ports", () => {
const stdout =
"tcp 0 0 127.0.0.1:20128 1.2.3.4:5555 ESTABLISHED 596922/node\n";
diff --git a/tests/unit/sse-auth-codex-account-pool.test.ts b/tests/unit/sse-auth-codex-account-pool.test.ts
index 581efd1f339..927775d15f7 100644
--- a/tests/unit/sse-auth-codex-account-pool.test.ts
+++ b/tests/unit/sse-auth-codex-account-pool.test.ts
@@ -242,8 +242,8 @@ test("Codex parent authentication failures block both virtual children without c
const inventory = await providersDb.getProviderConnections({ provider: "codex" });
assert.equal(unavailable.shouldFallback, true);
- assert.equal(spark, null);
- assert.equal(normal, null);
+ assert.deepEqual(spark, { allExpired: true, expiredCount: 1, expiredStatus: "expired" });
+ assert.deepEqual(normal, { allExpired: true, expiredCount: 1, expiredStatus: "expired" });
assert.deepEqual(
inventory.map((item) => item.id),
[connection.id]
diff --git a/tests/unit/stream-continuation-wiring.test.ts b/tests/unit/stream-continuation-wiring.test.ts
index 2f246e325d4..36a6f445d1f 100644
--- a/tests/unit/stream-continuation-wiring.test.ts
+++ b/tests/unit/stream-continuation-wiring.test.ts
@@ -47,9 +47,15 @@ async function collectText(stream: ReadableStream): Promise
const ROLE = 'data: {"choices":[{"delta":{"role":"assistant"}}]}\n\n';
const content = (s: string) => `data: {"choices":[{"delta":{"content":${JSON.stringify(s)}}}]}\n\n`;
+const reasoning = (s: string) =>
+ `data: {"choices":[{"delta":{"reasoning_content":${JSON.stringify(s)}}}]}\n\n`;
+const finishStopNoContent = 'data: {"choices":[{"delta":{},"finish_reason":"stop"}]}\n\n';
+const finishLengthNoContent = 'data: {"choices":[{"delta":{},"finish_reason":"length"}]}\n\n';
+
test("mid-stream continuation: stitches the suffix after a silent post-commit truncation", async () => {
- // Commits on chunk 1, emits "Hello wor", then ends WITHOUT a terminal marker (silent cut).
- const initial = streamFrom([ROLE, content("Hello wor")]);
+ // Commits on chunk 1, emits "Hello there world", then ends WITHOUT a terminal marker
+ // (silent cut).
+ const initial = streamFrom([ROLE, content("Hello there world")]);
let finalizeCount = 0;
let continueArg = "";
@@ -60,15 +66,22 @@ test("mid-stream continuation: stitches the suffix after a silent post-commit tr
now: steppingClock(),
continueStream: async (soFar: string) => {
continueArg = soFar;
- // The model re-emits a small overlap ("wor") which must be trimmed away.
- return streamFrom([ROLE, content("world!"), "data: [DONE]\n\n"]);
+ // The model re-emits only a partial tail of what was already sent ("there world",
+ // 11 chars — above the 8-char threshold, but NOT the full emitted text, unlike a
+ // full-string overlap this stays a discriminating test of trimContinuationOverlap's
+ // partial-tail trim, not just its "accept everything" path) before continuing.
+ return streamFrom([ROLE, content("there world, nice to meet you!"), "data: [DONE]\n\n"]);
},
});
const out = await collectText(stream);
const scan = scanOpenAiSseText(out);
- assert.equal(continueArg, "Hello wor", "continuation is prefilled with the text already sent");
- assert.equal(scan.text, "Hello world!", "client sees the full answer, overlap trimmed, exactly once");
+ assert.equal(continueArg, "Hello there world", "continuation is prefilled with the text already sent");
+ assert.equal(
+ scan.text,
+ "Hello there world, nice to meet you!",
+ "client sees the full answer, partial overlap trimmed, exactly once"
+ );
assert.equal(scan.terminal, true, "the recovered stream ends with a terminal marker");
assert.equal(finalizeCount, 1, "finalize runs exactly once");
});
@@ -80,7 +93,7 @@ test("mid-stream continuation: recovers a post-commit transport error too", asyn
const stream = createRecoverableStream(initial, async () => null, {
finalize: () => {},
now: steppingClock(),
- continueStream: async () => streamFrom([content("answer done."), "data: [DONE]\n\n"]),
+ continueStream: async () => streamFrom([content("Partial answer done."), "data: [DONE]\n\n"]),
});
const scan = scanOpenAiSseText(await collectText(stream));
assert.equal(scan.text, "Partial answer done.");
@@ -115,3 +128,167 @@ test("tool-call in flight is never continued (would corrupt tool JSON)", async (
await collectText(stream);
assert.equal(continued, false, "continuation must NOT fire once a tool call has started streaming");
});
+
+test("mid-stream continuation: a zero-overlap restart is rejected, never concatenated raw", async () => {
+ // Truncates silently after real, non-empty text — canContinue() fires.
+ const initial = streamFrom([ROLE, content("Tous les faits sont reunis")]);
+ let continuations = 0;
+ const stream = createRecoverableStream(initial, async () => null, {
+ finalize: () => {},
+ now: steppingClock(),
+ maxContinuations: 1,
+ continueStream: async () => {
+ continuations += 1;
+ // The model ignores the assistant prefill and restarts on an unrelated sentence —
+ // zero characters of overlap with what was already emitted.
+ return streamFrom([
+ content("Je complete le design - derniere verification"),
+ "data: [DONE]\n\n",
+ ]);
+ },
+ });
+ const out = await collectText(stream);
+ const scan = scanOpenAiSseText(out);
+ assert.equal(
+ scan.text,
+ "Tous les faits sont reunis",
+ "the unrelated restart must never be appended to the already-emitted text"
+ );
+ assert.equal(scan.terminal, true, "closes cleanly instead of leaving the client hanging");
+ assert.equal(continuations, 1, "bounded by maxContinuations — does not loop forever");
+});
+
+test("mid-stream continuation: a nonzero overlap below the threshold is rejected too", async () => {
+ // Genuine 4-character overlap ("pret"), well under the 8-char threshold — this is the
+ // false-negative case a naive `overlapChars === 0` check would miss (a restart that
+ // happens to share a short accidental fragment with the emitted tail): must still be
+ // treated as a suspected restart, not accepted as a genuine resume.
+ const initial = streamFrom([ROLE, content("Le design est pret")]);
+ let continuations = 0;
+ const stream = createRecoverableStream(initial, async () => null, {
+ finalize: () => {},
+ now: steppingClock(),
+ maxContinuations: 1,
+ continueStream: async () => {
+ continuations += 1;
+ // Shares only "pret" (4 chars) with the emitted tail, then diverges completely.
+ return streamFrom([content("pret a partir de zero"), "data: [DONE]\n\n"]);
+ },
+ });
+ const scan = scanOpenAiSseText(await collectText(stream));
+ assert.equal(
+ scan.text,
+ "Le design est pret",
+ "a below-threshold (but nonzero) overlap must not be accepted as a real resume"
+ );
+ assert.equal(continuations, 1);
+});
+
+test("mid-stream continuation: a real overlap at or above the threshold is still stitched correctly", async () => {
+ // Regression guard: the existing happy path (first test in this file, whose updated
+ // fixture re-emits the 11-char partial tail "there world") still passes below — this test
+ // adds an overlap AT the threshold boundary to prove Task 3's new check does not fire when
+ // it shouldn't.
+ const initial = streamFrom([ROLE, content("The answer to this question")]);
+ const stream = createRecoverableStream(initial, async () => null, {
+ finalize: () => {},
+ now: steppingClock(),
+ continueStream: async () =>
+ // "question" (8 chars) overlaps the tail of emittedText exactly at the threshold.
+ streamFrom([content("question is forty-two."), "data: [DONE]\n\n"]),
+ });
+ const scan = scanOpenAiSseText(await collectText(stream));
+ assert.equal(
+ scan.text,
+ "The answer to this question is forty-two.",
+ "an overlap meeting the threshold is trimmed and stitched, not rejected"
+ );
+});
+
+test("mid-stream continuation: a clean stop with reasoning-only output (no answer) triggers a continuation", async () => {
+ const initial = streamFrom([
+ ROLE,
+ reasoning("the model thinks through the problem here..."),
+ finishStopNoContent,
+ ]);
+ let continueArg = "__unset__";
+ const stream = createRecoverableStream(initial, async () => null, {
+ finalize: () => {},
+ now: steppingClock(),
+ continueStream: async (soFar: string) => {
+ continueArg = soFar;
+ return streamFrom([content("Here is the actual answer."), "data: [DONE]\n\n"]);
+ },
+ });
+ const out = await collectText(stream);
+ const scan = scanOpenAiSseText(out);
+ assert.equal(continueArg, "", "nothing usable was emitted — the re-request has an empty prefill");
+ assert.equal(
+ scan.text,
+ "Here is the actual answer.",
+ "the client gets a real answer instead of silence"
+ );
+ assert.equal(scan.terminal, true);
+});
+
+test("mid-stream continuation: a clean stop with truly empty output (no text, no reasoning) is left unchanged", async () => {
+ const initial = streamFrom([ROLE, finishStopNoContent]);
+ let continued = false;
+ const stream = createRecoverableStream(initial, async () => null, {
+ finalize: () => {},
+ now: steppingClock(),
+ continueStream: async () => {
+ continued = true;
+ return streamFrom([content("nope"), "data: [DONE]\n\n"]);
+ },
+ });
+ await collectText(stream);
+ assert.equal(
+ continued,
+ false,
+ "no reasoning trace means there is nothing to act on — do not guess"
+ );
+});
+
+test("mid-stream continuation: finish_reason 'length' with reasoning-only output does NOT trigger a continuation", async () => {
+ // Regression guard for a blocker found in cross-review: widening the gate to any
+ // terminal marker (instead of the literal finish_reason "stop") would wrongly spend a
+ // continuation attempt on a token-limit cutoff, which is out of this fix's scope.
+ const initial = streamFrom([
+ ROLE,
+ reasoning("the model was still thinking when it hit the token limit..."),
+ finishLengthNoContent,
+ ]);
+ let continued = false;
+ const stream = createRecoverableStream(initial, async () => null, {
+ finalize: () => {},
+ now: steppingClock(),
+ continueStream: async () => {
+ continued = true;
+ return streamFrom([content("nope"), "data: [DONE]\n\n"]);
+ },
+ });
+ await collectText(stream);
+ assert.equal(continued, false, "finish_reason 'length' is out of scope for this fix");
+});
+
+test("mid-stream continuation: real content alongside reasoning at a clean stop is left unchanged (non-regression)", async () => {
+ const initial = streamFrom([
+ ROLE,
+ reasoning("thinking..."),
+ content("The real answer."),
+ finishStopNoContent,
+ ]);
+ let continued = false;
+ const stream = createRecoverableStream(initial, async () => null, {
+ finalize: () => {},
+ now: steppingClock(),
+ continueStream: async () => {
+ continued = true;
+ return streamFrom([content("nope"), "data: [DONE]\n\n"]);
+ },
+ });
+ const scan = scanOpenAiSseText(await collectText(stream));
+ assert.equal(continued, false, "real content was delivered — nothing to recover");
+ assert.equal(scan.text, "The real answer.");
+});
diff --git a/tests/unit/stream-continuation.test.ts b/tests/unit/stream-continuation.test.ts
index caffe936df9..8a1da37ce9d 100644
--- a/tests/unit/stream-continuation.test.ts
+++ b/tests/unit/stream-continuation.test.ts
@@ -21,6 +21,30 @@ test("scanOpenAiSseText accumulates content deltas and flags an OpenAI-compat st
assert.equal(r.terminal, false);
});
+test("scanOpenAiSseText accumulates reasoning_content deltas separately from content", () => {
+ const sse =
+ 'data: {"choices":[{"delta":{"role":"assistant"}}]}\n\n' +
+ 'data: {"choices":[{"delta":{"reasoning_content":"thinking..."}}]}\n\n' +
+ 'data: {"choices":[{"delta":{"reasoning_content":" more"}}]}\n\n';
+ const r = scanOpenAiSseText(sse);
+ assert.equal(r.reasoningText, "thinking... more");
+ assert.equal(r.text, "", "reasoning_content must never leak into the visible text field");
+ assert.equal(r.parsedOpenAi, true);
+});
+
+test("scanOpenAiSseText captures the literal finish_reason value", () => {
+ const stop = scanOpenAiSseText('data: {"choices":[{"delta":{},"finish_reason":"stop"}]}\n\n');
+ assert.equal(stop.finishReason, "stop");
+
+ const length = scanOpenAiSseText(
+ 'data: {"choices":[{"delta":{"content":"x"},"finish_reason":"length"}]}\n\n'
+ );
+ assert.equal(length.finishReason, "length");
+
+ const none = scanOpenAiSseText('data: {"choices":[{"delta":{"content":"x"}}]}\n\n');
+ assert.equal(none.finishReason, null, "no finish_reason seen means null, not a guessed default");
+});
+
test("scanOpenAiSseText detects the terminal [DONE] marker", () => {
const r = scanOpenAiSseText('data: {"choices":[{"delta":{"content":"hi"}}]}\n\ndata: [DONE]\n\n');
assert.equal(r.text, "hi");
@@ -65,6 +89,15 @@ test("makeContinuationBody refuses bodies without a messages array or empty text
assert.equal(makeContinuationBody(null as never, "t"), null);
});
+test("makeContinuationBody accepts an empty prefill by re-sending the messages unchanged", () => {
+ const body = { model: "x", stream: true, messages: [{ role: "user", content: "hi" }] };
+ const out = makeContinuationBody(body, "");
+ assert.ok(out, "an empty prefill must still produce a re-request body, not null");
+ assert.equal(out!.messages.length, 1, "no empty assistant turn is appended");
+ assert.deepEqual(out!.messages[0], { role: "user", content: "hi" });
+ assert.equal(out!.stream, true);
+});
+
// ── trimContinuationOverlap ───────────────────────────────────────────────────
test("trimContinuationOverlap removes a duplicated seam so the join is append-only", () => {
diff --git a/tests/unit/stream-handler.test.ts b/tests/unit/stream-handler.test.ts
index d19a13f351e..8c19c088021 100644
--- a/tests/unit/stream-handler.test.ts
+++ b/tests/unit/stream-handler.test.ts
@@ -167,6 +167,41 @@ test("createDisconnectAwareStream treats cancel after Responses completed as suc
assert.equal(disconnectHandled, false);
});
+test("createDisconnectAwareStream recognizes a large Responses compaction completion", async () => {
+ let errorHandled = false;
+ const completed = `event: response.completed\ndata: ${JSON.stringify({
+ type: "response.completed",
+ response: {
+ status: "completed",
+ output: [{ type: "compaction", encrypted_content: "x".repeat(5000) }],
+ },
+ })}\n\n`;
+ const transformStream = {
+ readable: new ReadableStream({
+ start(controller) {
+ controller.enqueue(encoder.encode(completed));
+ controller.close();
+ },
+ }),
+ writable: createNoopAbortWritable(),
+ };
+
+ const stream = createDisconnectAwareStream(
+ transformStream,
+ createStreamController({
+ clientResponseFormat: FORMATS.OPENAI_RESPONSES,
+ onError() {
+ errorHandled = true;
+ },
+ })
+ );
+ const text = await readStreamText(stream);
+
+ assert.equal(text, completed);
+ assert.equal(errorHandled, false);
+ assert.doesNotMatch(text, /response\.failed/);
+});
+
test("createDisconnectAwareStream: Gemini 503 high-demand error becomes SSE error chunk with message preserved", async () => {
const geminiMsg =
"[503]: This model is currently experiencing high demand. Spikes in demand are usually temporary. Please try again later.";
diff --git a/tests/unit/strip-reasoning-blobs-agentic-context-1599.test.ts b/tests/unit/strip-reasoning-blobs-agentic-context-1599.test.ts
index 25eae7a839c..8b3059236dd 100644
--- a/tests/unit/strip-reasoning-blobs-agentic-context-1599.test.ts
+++ b/tests/unit/strip-reasoning-blobs-agentic-context-1599.test.ts
@@ -8,7 +8,7 @@ import { omitEncryptedReasoningForLog } from "../../src/lib/logPayloads.ts";
// Responses reasoning replay is target-scoped. Plaintext DeepSeek state and
// provider-generated opaque state are never interchangeable.
-test("unknown Responses targets reject opaque reasoning and ignore display summaries", () => {
+test("unknown Responses targets drop opaque reasoning and preserve display summaries (#10959)", () => {
const body: Record = {
input: [
{ type: "message", role: "user", content: [{ type: "input_text", text: "hi" }] },
@@ -24,11 +24,14 @@ test("unknown Responses targets reject opaque reasoning and ignore display summa
],
};
- const originalInput = structuredClone(body.input);
const result = applyReasoningInputPolicy(body, "responses");
- assert.equal(result.incompatibleReasoning, true);
- assert.deepEqual(body.input, originalInput, "rejection must not mutate the request");
+ assert.equal(result.incompatibleReasoning, false);
+ assert.deepEqual(body.input, [
+ { type: "message", role: "user", content: [{ type: "input_text", text: "hi" }] },
+ { type: "reasoning", summary: [{ text: "display only" }] },
+ { type: "function_call", name: "search", arguments: "{}", call_id: "call_1" },
+ ]);
});
test("unannotated targets preserve plaintext Responses reasoning without synthetic IDs", () => {
@@ -132,7 +135,7 @@ test("Chat drop removes opaque state while preserving plaintext and summary deta
]);
});
-test("DeepSeek rejects plaintext reasoning carrying opaque provider state", () => {
+test("DeepSeek projects plaintext reasoning carrying opaque provider state onto the plaintext transport (#10949)", () => {
for (const opaqueField of ["signature", "format"] as const) {
const body: Record = {
input: [
@@ -148,13 +151,11 @@ test("DeepSeek rejects plaintext reasoning carrying opaque provider state", () =
const result = applyReasoningInputPolicy(body, "responses", { provider: "deepseek" });
- assert.equal(result.incompatibleReasoning, true, opaqueField);
+ assert.equal(result.incompatibleReasoning, false, opaqueField);
assert.deepEqual(body.input, [
{
- id: "rs_mixed123",
type: "reasoning",
content: [{ type: "reasoning_text", text: "untrusted companion" }],
- [opaqueField]: "provider-state",
},
{ type: "message", role: "user", content: [{ type: "input_text", text: "continue" }] },
]);
@@ -199,6 +200,49 @@ test("drop fallback removes only the incompatible active transport and preserves
assert.equal(opaqueReasoning.encrypted_content, "provider-state");
});
+test("mixed plaintext + opaque reasoning follows the target transport instead of rejecting (#10949)", () => {
+ const mixedReasoning = {
+ id: "rs_mixed",
+ type: "reasoning",
+ content: [{ type: "reasoning_text", text: "inspect first" }],
+ encrypted_content: "provider-state",
+ summary: [{ type: "summary_text", text: "display only" }],
+ };
+
+ // Plaintext target (deepseek): keep the portable plaintext, strip opaque state.
+ const toPlaintext: Record = {
+ input: [structuredClone(mixedReasoning)],
+ };
+ const plaintextResult = applyReasoningInputPolicy(toPlaintext, "responses", {
+ provider: "deepseek",
+ });
+ assert.equal(plaintextResult.incompatibleReasoning, false);
+ assert.deepEqual(toPlaintext.input, [
+ {
+ type: "reasoning",
+ content: [{ type: "reasoning_text", text: "inspect first" }],
+ summary: [{ type: "summary_text", text: "display only" }],
+ },
+ ]);
+
+ // Opaque target (openai): keep the provider state, strip the plaintext.
+ const toOpaque: Record = {
+ input: [structuredClone(mixedReasoning)],
+ };
+ const opaqueResult = applyReasoningInputPolicy(toOpaque, "responses", {
+ provider: "openai",
+ });
+ assert.equal(opaqueResult.incompatibleReasoning, false);
+ assert.deepEqual(toOpaque.input, [
+ {
+ id: "rs_mixed",
+ type: "reasoning",
+ encrypted_content: "provider-state",
+ summary: [{ type: "summary_text", text: "display only" }],
+ },
+ ]);
+});
+
test("drop fallback preserves reasoning when its transport is compatible", () => {
const body: Record = {
input: [
diff --git a/tests/unit/systemd-notify.test.mjs b/tests/unit/systemd-notify.test.mjs
index cb9e441f8f7..b716d57d9d5 100644
--- a/tests/unit/systemd-notify.test.mjs
+++ b/tests/unit/systemd-notify.test.mjs
@@ -37,8 +37,10 @@ function python3Available() {
}
}
-// Waits for the listener to emit `expected` lines (in order), then resolves
-// with everything it saw. Fails loudly on timeout or premature exit.
+// Waits for the listener to emit every `expected` line, then resolves with
+// everything it saw. The notifier spawns one process per signal, so AF_UNIX
+// datagram arrival order is not guaranteed across those processes.
+// Fails loudly on timeout or premature exit.
// BARRIER=1 datagrams (sd_notify synchronization emitted by the systemd-notify
// CLI after every message) are noise for this contract and are skipped.
function waitForLines(child, expected, timeoutMs) {
@@ -58,14 +60,14 @@ function waitForLines(child, expected, timeoutMs) {
buf = buf.slice(idx + 1);
if (!line || line === "BARRIER=1") continue;
seen.push(line);
- if (seen.length === expected.length) {
+ if (expected.every((expectedLine) => seen.includes(expectedLine))) {
clearTimeout(timer);
resolve([...seen]);
}
}
});
child.on("exit", () => {
- if (seen.length < expected.length) {
+ if (expected.some((expectedLine) => !seen.includes(expectedLine))) {
clearTimeout(timer);
reject(new Error(`listener exited early; got: ${seen.join(", ")}`));
}
@@ -248,7 +250,7 @@ test(
notifier.watchdog();
notifier.stopping();
const received = await waitForLines(listener, ["READY=1", "WATCHDOG=1", "STOPPING=1"], 10000);
- assert.deepEqual(received, ["READY=1", "WATCHDOG=1", "STOPPING=1"]);
+ assert.deepEqual(received.toSorted(), ["READY=1", "STOPPING=1", "WATCHDOG=1"]);
notifier.dispose();
} finally {
listener.kill();
diff --git a/tests/unit/terminal-status-origin.test.ts b/tests/unit/terminal-status-origin.test.ts
index a4aacf767fc..3a9221f2cab 100644
--- a/tests/unit/terminal-status-origin.test.ts
+++ b/tests/unit/terminal-status-origin.test.ts
@@ -1,6 +1,8 @@
import { test } from "node:test";
import assert from "node:assert/strict";
-import fs from "node:fs"; import os from "node:os"; import path from "node:path";
+import fs from "node:fs";
+import os from "node:os";
+import path from "node:path";
const DIR = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-t11-"));
process.env.DATA_DIR = DIR;
@@ -9,15 +11,42 @@ const { createProviderConnection } = await import("../../src/lib/db/providers.ts
const { runAsProbe } = await import("../../src/shared/utils/probeOrigin.ts");
const { writeTerminalStatus } = await import("../../src/shared/utils/terminalStatus.ts");
-test.after(() => { core.resetDbInstance(); fs.rmSync(DIR, {recursive:true, force:true}); });
+test.after(() => {
+ core.resetDbInstance();
+ fs.rmSync(DIR, { recursive: true, force: true });
+});
-function row(id: string){ return (core.getDbInstance() as unknown as Record).prepare("SELECT is_active, test_status FROM provider_connections WHERE id=?").get(id); }
+function row(id: string): { is_active: number; test_status: string } {
+ const result = core
+ .getDbInstance()
+ .prepare("SELECT is_active, test_status FROM provider_connections WHERE id=?")
+ .get(id);
+ assert.ok(result && typeof result === "object");
+ return result as { is_active: number; test_status: string };
+}
test("probe-origin writeTerminalStatus records error but never deactivates", async () => {
- const conn = await createProviderConnection({ provider:"openai", authType:"apikey", name:"t11", apiKey:"sk-t11", isActive:true, testStatus:"active" } as unknown as Record);
- const id = String((conn as unknown as Record).id);
+ const conn = await createProviderConnection({
+ provider: "openai",
+ authType: "apikey",
+ name: "t11",
+ apiKey: "sk-t11",
+ isActive: true,
+ testStatus: "active",
+ });
+ const id = String(conn.id);
await runAsProbe(async () => {
- await writeTerminalStatus(id, { testStatus:"banned", isActive:false, lastError:"probe 403", errorCode:"403", lastErrorType:"FORBIDDEN" }, "probe");
+ await writeTerminalStatus(
+ id,
+ {
+ testStatus: "banned",
+ isActive: false,
+ lastError: "probe 403",
+ errorCode: "403",
+ lastErrorType: "FORBIDDEN",
+ },
+ "probe"
+ );
});
const r = row(id);
assert.equal(r.is_active, 1); // probe n'a jamais désactivé
@@ -25,9 +54,26 @@ test("probe-origin writeTerminalStatus records error but never deactivates", asy
});
test("production writeTerminalStatus deactivates on terminal", async () => {
- const conn = await createProviderConnection({ provider:"openai", authType:"apikey", name:"t11b", apiKey:"sk-t11b", isActive:true, testStatus:"active" } as unknown as Record);
- const id = String((conn as unknown as Record).id);
- await writeTerminalStatus(id, { testStatus:"banned", isActive:false, lastError:"real 403", errorCode:"403", lastErrorType:"FORBIDDEN" }, "production");
+ const conn = await createProviderConnection({
+ provider: "openai",
+ authType: "apikey",
+ name: "t11b",
+ apiKey: "sk-t11b",
+ isActive: true,
+ testStatus: "active",
+ });
+ const id = String(conn.id);
+ await writeTerminalStatus(
+ id,
+ {
+ testStatus: "banned",
+ isActive: false,
+ lastError: "real 403",
+ errorCode: "403",
+ lastErrorType: "FORBIDDEN",
+ },
+ "production"
+ );
const r = row(id);
assert.equal(r.is_active, 0);
assert.equal(r.test_status, "banned");
diff --git a/tests/unit/translator-openai-responses-req.test.ts b/tests/unit/translator-openai-responses-req.test.ts
index da87b26116f..2a05f57310e 100644
--- a/tests/unit/translator-openai-responses-req.test.ts
+++ b/tests/unit/translator-openai-responses-req.test.ts
@@ -172,28 +172,26 @@ test("Responses -> Chat keeps summary-only reasoning out of continuation state",
assert.equal(result.messages[0].reasoning_content, undefined);
});
-test("Responses -> Chat rejects opaque reasoning instead of replaying its plaintext companion", () => {
- assert.throws(
- () =>
- openaiResponsesToOpenAIRequest(
- "deepseek-v4-pro",
+test("Responses -> Chat replays the plaintext companion of an opaque reasoning item (#10949)", () => {
+ const result = openaiResponsesToOpenAIRequest(
+ "deepseek-v4-pro",
+ {
+ input: [
{
- input: [
- {
- id: "rs_opaque",
- type: "reasoning",
- encrypted_content: "opaque-provider-state",
- content: [{ type: "reasoning_text", text: "Untrusted plaintext companion" }],
- summary: [{ type: "summary_text", text: "Display summary" }],
- },
- { type: "function_call", call_id: "call_1", name: "search", arguments: "{}" },
- ],
+ id: "rs_opaque",
+ type: "reasoning",
+ encrypted_content: "opaque-provider-state",
+ content: [{ type: "reasoning_text", text: "Untrusted plaintext companion" }],
+ summary: [{ type: "summary_text", text: "Display summary" }],
},
- false,
- { _preserveReasoningContent: true }
- ),
- /Reasoning continuation is not compatible/
- );
+ { type: "function_call", call_id: "call_1", name: "search", arguments: "{}" },
+ ],
+ },
+ false,
+ { _preserveReasoningContent: true }
+ ) as { messages: Array> };
+
+ assert.equal(result.messages[0].reasoning_content, "Untrusted plaintext companion");
});
test("Responses -> Chat merges assistant text that follows a function call", () => {
diff --git a/tests/unit/translator-resp-openai-responses.test.ts b/tests/unit/translator-resp-openai-responses.test.ts
index 27cc877f2ef..2253c655ac7 100644
--- a/tests/unit/translator-resp-openai-responses.test.ts
+++ b/tests/unit/translator-resp-openai-responses.test.ts
@@ -423,6 +423,52 @@ test("Responses -> OpenAI: preserves non-object Read JSON-string arguments", ()
assert.equal(done.choices[0].delta.tool_calls[0].function.arguments, "null");
});
+test("Responses -> OpenAI: mixed plaintext + encrypted_content reasoning replays its plaintext (#10949)", () => {
+ const state = {};
+ const done = openaiResponsesToOpenAIResponse(
+ {
+ type: "response.output_item.done",
+ item: {
+ type: "reasoning",
+ id: "rs_mixed",
+ content: [
+ {
+ type: "reasoning_text",
+ text: "Let me start by reading the directory to understand the structure of the corpus.",
+ },
+ ],
+ encrypted_content: "",
+ summary: [],
+ },
+ },
+ state
+ );
+
+ assert.ok(done, "mixed reasoning item must surface a delta");
+ assert.equal(
+ done.choices[0].delta.reasoning_content,
+ "Let me start by reading the directory to understand the structure of the corpus."
+ );
+});
+
+test("Responses -> OpenAI: opaque-only reasoning still emits no fabricated plaintext", () => {
+ const state = {};
+ const done = openaiResponsesToOpenAIResponse(
+ {
+ type: "response.output_item.done",
+ item: {
+ type: "reasoning",
+ id: "rs_opaque_only",
+ encrypted_content: "",
+ summary: [],
+ },
+ },
+ state
+ );
+
+ assert.equal(done, null);
+});
+
test("Responses -> OpenAI: strips empty optional args from JSON-string output_item.done arguments", () => {
const state = {};
openaiResponsesToOpenAIResponse(
diff --git a/tests/unit/ui/cheaperInferenceSponsorBanner.test.tsx b/tests/unit/ui/cheaperInferenceSponsorBanner.test.tsx
new file mode 100644
index 00000000000..b86f4188c6a
--- /dev/null
+++ b/tests/unit/ui/cheaperInferenceSponsorBanner.test.tsx
@@ -0,0 +1,92 @@
+// @vitest-environment jsdom
+/**
+ * CheaperInferenceSponsorBanner — render gate (localStorage dismissal), CTA
+ * pointing at our link.omniroute.online branded short link, and discreet
+ * partner-link note. Mirrors kimiSponsorBanner.test.tsx, minus the version gate
+ * (this banner is a durable partnership, not a time-boxed offer).
+ */
+import React from "react";
+import { act } from "react";
+import { createRoot } from "react-dom/client";
+import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
+
+const STORAGE_KEY = "omniroute-cheaperinference-sponsor-banner-dismissed-v1";
+const DISMISS_EVENT = "omniroute:cheaperinference-sponsor-banner-dismissed";
+const SHORT_URL = "https://link.omniroute.online/cheaper";
+
+vi.mock("next-intl", () => ({ useTranslations: () => (k: string) => k }));
+vi.mock("@/shared/components/ProviderIcon", () => ({ default: () => null }));
+
+async function renderBanner(): Promise {
+ vi.resetModules();
+ const { default: CheaperInferenceSponsorBanner } =
+ await import("../../../src/app/(dashboard)/dashboard/CheaperInferenceSponsorBanner");
+
+ const container = document.createElement("div");
+ document.body.appendChild(container);
+ const root = createRoot(container);
+ act(() => {
+ root.render();
+ });
+ return container;
+}
+
+describe("CheaperInferenceSponsorBanner", () => {
+ beforeEach(() => {
+ (
+ globalThis as typeof globalThis & { IS_REACT_ACT_ENVIRONMENT?: boolean }
+ ).IS_REACT_ACT_ENVIRONMENT = true;
+ localStorage.removeItem(STORAGE_KEY);
+ });
+
+ afterEach(() => {
+ document.body.innerHTML = "";
+ localStorage.removeItem(STORAGE_KEY);
+ });
+
+ it("renders with the CTA pointing at the branded short link", async () => {
+ const container = await renderBanner();
+ expect(container.textContent).toContain("title");
+ expect(container.textContent).toContain("cta");
+ const link = container.querySelector("a[href]");
+ expect(link).not.toBeNull();
+ expect(link?.getAttribute("href")).toBe(SHORT_URL);
+ expect(link?.getAttribute("target")).toBe("_blank");
+ expect(link?.getAttribute("rel")).toContain("noopener");
+ });
+
+ it("shows the discreet partner-link note near the CTA", async () => {
+ const container = await renderBanner();
+ expect(container.textContent).toContain("partnerLinkNote");
+ const link = container.querySelector("a[href]");
+ expect(link?.getAttribute("title")).toBe("partnerLinkNote");
+ });
+
+ it("hides after dismissal and stays hidden on re-render", async () => {
+ const first = await renderBanner();
+ const button = first.querySelector("button");
+ expect(button).not.toBeNull();
+ act(() => {
+ button?.click();
+ });
+ expect(localStorage.getItem(STORAGE_KEY)).toBe("true");
+ expect(first.textContent).not.toContain("title");
+
+ // a fresh render (simulating a later visit) stays hidden
+ const second = await renderBanner();
+ expect(second.textContent).not.toContain("title");
+ });
+
+ it("re-renders visible again only after the key is cleared", async () => {
+ const first = await renderBanner();
+ const button = first.querySelector("button");
+ act(() => {
+ button?.click();
+ });
+ expect(localStorage.getItem(STORAGE_KEY)).toBe("true");
+
+ localStorage.removeItem(STORAGE_KEY);
+ const second = await renderBanner();
+ expect(second.textContent).toContain("title");
+ });
+});
diff --git a/tests/unit/ui/vscodeCopilotBanner.test.tsx b/tests/unit/ui/vscodeCopilotBanner.test.tsx
index ed75ad4950a..fad50b2b266 100644
--- a/tests/unit/ui/vscodeCopilotBanner.test.tsx
+++ b/tests/unit/ui/vscodeCopilotBanner.test.tsx
@@ -11,7 +11,7 @@ import { createRoot } from "react-dom/client";
import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
const STORAGE_KEY = "omniroute-vscode-copilot-banner-dismissed-v1";
-const MARKETPLACE_URL = "https://marketplace.visualstudio.com/items?itemName=diegosouzapw.omnicopilot";
+const MARKETPLACE_URL = "https://link.omniroute.online/vsx";
vi.mock("next-intl", () => ({ useTranslations: () => (k: string) => k }));
diff --git a/tests/unit/usage-command-json-format.test.ts b/tests/unit/usage-command-json-format.test.ts
new file mode 100644
index 00000000000..9f76d6e1913
--- /dev/null
+++ b/tests/unit/usage-command-json-format.test.ts
@@ -0,0 +1,162 @@
+/**
+ * #8 (OmniCopilot) — the usage command answered `text/plain`, which a UI cannot
+ * parse safely. The structured form (`?format=json`) returns the same
+ * `ApiKeyUsageLimitStatus` + `UsageSnapshot` the text is rendered from.
+ *
+ * These tests pin the contract the extension depends on: JSON when asked,
+ * text by default, the 403 as a structured reason rather than a bare string,
+ * and an error body that never carries a stack trace.
+ */
+import test from "node:test";
+import assert from "node:assert/strict";
+
+import { handleInternalUsageCommandHttpRequest } from "../../src/lib/usage/internalUsageCommand";
+
+const NOW = Date.parse("2026-08-19T12:00:00.000Z");
+
+const LIMIT_STATUS = {
+ enabled: true,
+ dailyLimitUsd: 5,
+ weeklyLimitUsd: 20,
+ dailySpentUsd: 1.25,
+ weeklySpentUsd: 8,
+ dailyWindowStartIso: "2026-08-19T03:00:00.000Z",
+ dailyResetAtIso: "2026-08-20T03:00:00.000Z",
+ weeklyWindowStartIso: "2026-08-16T03:00:00.000Z",
+ weeklyResetAtIso: "2026-08-23T03:00:00.000Z",
+ dailyExceeded: false,
+ weeklyExceeded: false,
+};
+
+function allowedDeps(overrides: Record = {}) {
+ return {
+ now: () => NOW,
+ isValidApiKey: async (apiKey: string) => apiKey === "sk-allowed",
+ getApiKeyMetadata: async () => ({
+ id: "key-allowed",
+ name: "panel key",
+ allowUsageCommand: true,
+ usageLimitEnabled: true,
+ }),
+ getProviderConnections: async () => [
+ { id: "conn-claude", provider: "claude", isActive: true },
+ { id: "conn-codex", provider: "codex", isActive: true },
+ ],
+ getAllProviderLimitsCache: () => ({
+ "conn-claude": {
+ plan: "Claude Max",
+ quotas: {
+ weekly: { used: 25, total: 100, remaining: 75, resetAt: "2026-08-25T03:00:00.000Z" },
+ },
+ message: null,
+ fetchedAt: new Date(NOW).toISOString(),
+ },
+ "conn-codex": {
+ plan: "Codex Pro",
+ quotas: {
+ weekly: { used: 9, total: 100, remaining: 91, resetAt: "2026-08-24T03:00:00.000Z" },
+ },
+ message: null,
+ fetchedAt: new Date(NOW).toISOString(),
+ },
+ }),
+ getProviderConnectionById: async () => null,
+ getProviderLimitsCache: () => null,
+ getQuotaPolicy: async () => ({ defaultThresholdPercent: 0, providerWindowDefaults: {} }),
+ getApiKeyUsageLimitStatus: async () => LIMIT_STATUS,
+ ...overrides,
+ };
+}
+
+test("om-usage ?format=json returns the structured personal + provider quota", async () => {
+ const response = await handleInternalUsageCommandHttpRequest(
+ new Request("http://localhost/api/usage/om-usage?format=json", {
+ headers: { Authorization: "Bearer sk-allowed" },
+ }),
+ allowedDeps()
+ );
+
+ assert.equal(response.status, 200);
+ assert.match(response.headers.get("content-type") ?? "", /application\/json/);
+ const body = (await response.json()) as {
+ allowed: boolean;
+ personal: { dailySpentUsd: number } | null;
+ provider: { provider: string; connectionId: string } | null;
+ };
+ assert.equal(body.allowed, true);
+ assert.equal(body.personal?.dailySpentUsd, 1.25);
+ assert.equal(body.provider?.provider, "claude");
+ assert.equal(body.provider?.connectionId, "conn-claude");
+});
+
+test("om-usage without ?format stays text/plain (the historical contract)", async () => {
+ const response = await handleInternalUsageCommandHttpRequest(
+ new Request("http://localhost/api/usage/om-usage", {
+ headers: { Authorization: "Bearer sk-allowed" },
+ }),
+ allowedDeps()
+ );
+
+ assert.equal(response.status, 200);
+ assert.match(response.headers.get("content-type") ?? "", /text\/plain/);
+ const text = await response.text();
+ assert.match(text, /Personal quota/);
+ assert.match(text, /Provider quota/);
+});
+
+test("om-usage ?format=json returns every connection under providers[], not just the selected one", async () => {
+ // #11191 — a panel needs Codex + Claude side by side; the single `provider`
+ // pick is a terminal presentation choice, the collector had them all.
+ const response = await handleInternalUsageCommandHttpRequest(
+ new Request("http://localhost/api/usage/om-usage?format=json", {
+ headers: { Authorization: "Bearer sk-allowed" },
+ }),
+ allowedDeps()
+ );
+
+ assert.equal(response.status, 200);
+ const body = (await response.json()) as {
+ allowed: boolean;
+ provider: { provider: string } | null;
+ providers: Array<{ provider: string }>;
+ };
+ assert.equal(body.allowed, true);
+ const names = body.providers.map((s) => s.provider).sort();
+ assert.deepEqual(names, ["claude", "codex"]);
+ // the single-pick field is still present and one of them
+ assert.ok(["claude", "codex"].includes(body.provider?.provider ?? ""));
+});
+
+test("om-usage ?format=json reports a disallowed key as structured allowed:false", async () => {
+ // A usage panel must tell "this key may not ask" apart from "no data yet",
+ // which a bare 403 text body cannot express.
+ const response = await handleInternalUsageCommandHttpRequest(
+ new Request("http://localhost/api/usage/om-usage?format=json", {
+ headers: { Authorization: "Bearer sk-allowed" },
+ }),
+ allowedDeps({
+ getApiKeyMetadata: async () => ({ id: "key-off", allowUsageCommand: false }),
+ })
+ );
+
+ assert.equal(response.status, 403);
+ const body = (await response.json()) as { allowed: boolean };
+ assert.equal(body.allowed, false);
+});
+
+test("om-usage ?format=json rejects an invalid key and never leaks a stack trace", async () => {
+ const response = await handleInternalUsageCommandHttpRequest(
+ new Request("http://localhost/api/usage/om-usage?format=json", {
+ headers: { Authorization: "Bearer sk-wrong" },
+ }),
+ allowedDeps()
+ );
+
+ assert.equal(response.status, 401);
+ const body = (await response.json()) as { allowed: boolean; error?: { message?: string } };
+ assert.equal(body.allowed, false);
+ assert.ok(
+ !body.error?.message?.includes("at /"),
+ "error bodies must not carry stack frames (ERROR_SANITIZATION)"
+ );
+});
diff --git a/tests/unit/validate-response-quality.test.ts b/tests/unit/validate-response-quality.test.ts
index de4b5f47dc3..6a918155e56 100644
--- a/tests/unit/validate-response-quality.test.ts
+++ b/tests/unit/validate-response-quality.test.ts
@@ -24,6 +24,20 @@ test("returns valid=true for SSE with 'data:' lines", async () => {
assert.strictEqual(res.valid, true);
});
+test("returns valid=true for SSE opening with a ':' comment line (e.g. OpenRouter keep-alive)", async () => {
+ const res = await validateResponseQuality(
+ makeResponse(': OPENROUTER PROCESSING\n\ndata: {"foo":"bar"}\n\n'),
+ false,
+ {}
+ );
+ assert.strictEqual(res.valid, true);
+});
+
+test("returns valid=true for an SSE stream with leading whitespace before the first frame", async () => {
+ const res = await validateResponseQuality(makeResponse('\n\ndata: {"foo":"bar"}\n\n'), false, {});
+ assert.strictEqual(res.valid, true);
+});
+
test("returns valid=false for non-JSON non-SSE text", async () => {
const res = await validateResponseQuality(makeResponse("Hello world"), false, {});
assert.strictEqual(res.valid, false);
diff --git a/tests/unit/vision-bridge-maxchars.test.ts b/tests/unit/vision-bridge-maxchars.test.ts
index d667e515621..7f364468c73 100644
--- a/tests/unit/vision-bridge-maxchars.test.ts
+++ b/tests/unit/vision-bridge-maxchars.test.ts
@@ -67,6 +67,11 @@ test("modalityBridgeVisionMaxChars=120 caps the description with a … suffix",
}),
callVisionModel: async (_imageDataUri: string, _config: VisionModelConfig) =>
LONG_DESCRIPTION,
+ // #10859 made the reroute heuristic try a live vision-capable model for
+ // not-combo text-only models, which would hijack the request before the
+ // describe path. Pin credentials to definitively-unusable (false) so the
+ // reroute is excluded and the describe path under test runs.
+ hasUsableCredentials: async () => false,
},
});
@@ -93,6 +98,8 @@ test("no modalityBridgeVisionMaxChars key: description is passed through in full
}),
callVisionModel: async (_imageDataUri: string, _config: VisionModelConfig) =>
LONG_DESCRIPTION,
+ // Same #10859 reroute guard as above — keep the describe path under test.
+ hasUsableCredentials: async () => false,
},
});
@@ -125,6 +132,8 @@ test("updateSettingsSchema accepts an explicit modalityBridgeVisionMaxChars: 0 t
}),
callVisionModel: async (_imageDataUri: string, _config: VisionModelConfig) =>
LONG_DESCRIPTION,
+ // Same #10859 reroute guard as above — keep the describe path under test.
+ hasUsableCredentials: async () => false,
},
});
diff --git a/tests/unit/zcode-executor.test.ts b/tests/unit/zcode-executor.test.ts
index c82e7b33683..db4a8586bee 100644
--- a/tests/unit/zcode-executor.test.ts
+++ b/tests/unit/zcode-executor.test.ts
@@ -26,6 +26,8 @@ function requestBody() {
test("ZCode accepts GLM Coding Plan models and rejects unsafe/unknown ids", async () => {
const { resolveZcodeModel } = await loadZcodeExecutor();
assert.deepEqual(resolveZcodeModel("glm-5.2"), { ok: true, model: "glm-5.2" });
+ assert.equal(resolveZcodeModel("glm-5.2-high").ok, false);
+ assert.equal(resolveZcodeModel("glm-5.3-low").ok, false);
assert.equal(resolveZcodeModel("-unexpected").ok, false);
assert.equal(resolveZcodeModel("unknown-model").ok, false);
});
@@ -70,7 +72,7 @@ test("ZCode buffers the completed turn into OpenAI SSE when stream=true", async
});
const result = await executor.execute({
- model: "glm-5.2-high",
+ model: "glm-5.2",
body: requestBody(),
stream: true,
credentials: {},
diff --git a/tests/unit/zcode-provider.test.ts b/tests/unit/zcode-provider.test.ts
index 3acf3c862c4..e882ab5fd91 100644
--- a/tests/unit/zcode-provider.test.ts
+++ b/tests/unit/zcode-provider.test.ts
@@ -10,5 +10,18 @@ test("ZCode provider registry exposes a local no-auth GLM Coding Plan backend",
assert.equal(zcodeProvider.baseUrl, "zcode://app-server/stdio");
assert.equal(zcodeProvider.authType, "none");
assert.equal(zcodeProvider.authHeader, "none");
- assert.equal(zcodeProvider.models.some((model) => model.id === "glm-5.2"), true);
+ assert.equal(
+ zcodeProvider.models.some((model) => model.id === "glm-5.2"),
+ true
+ );
+ for (const alias of ["glm-5.3-high", "glm-5.3-low", "glm-5.2-high", "glm-5.2-max"]) {
+ assert.equal(
+ zcodeProvider.models.some((model) => model.id === alias),
+ false,
+ alias
+ );
+ }
+ for (const model of zcodeProvider.models) {
+ assert.deepEqual(model.supportedThinkingEfforts, [], model.id);
+ }
});
+
+
+