feat(providers): add openai-discount provider

[steebchen]

Summary

Adds a new openai-discount provider that behaves exactly like upstream OpenAI for all request/response handling, mirroring the existing anthropic-discount pattern, with two differences:

• Base URL + API key come from environment variables (LLM_OPENAI_DISCOUNT_BASE_URL / LLM_OPENAI_DISCOUNT_API_KEY), following the same stealth pattern as anthropic-discount / glacier / quartz.
• Pricing carries a 10% discount relative to the upstream prices.

What changed

• packages/models: new openai-discount provider definition (env-based apiKey + baseUrl); the existing gpt-5.5 model gains an openai-discount provider mapping with discount: "0.1" and supportsResponsesApi: true.
• packages/actions: endpoint resolution reads LLM_OPENAI_DISCOUNT_BASE_URL and appends /v1/responses (or /v1/chat/completions); the request-body transform (Responses + chat completions paths) treats openai-discount identically to openai. The responses-API capability lookup now keys on the actual provider instead of a hardcoded "openai".
• apps/gateway: OpenAI-specific branches (Responses-API streaming/non-streaming transforms, tool-call extraction, usage extraction, web-search counting) now also match openai-discount.
• .env.example: documents the two new env vars.

Token, header (Bearer), and response parsing flow through the existing generic / OpenAI-default machinery, so no further plumbing was needed.

Testing

• pnpm format ✅
• pnpm build ✅
• pnpm test:unit ✅ (1696 passed)

End-to-end verification against the real discounted endpoint is pending — the proxy will be configured separately.

🤖 Generated with Claude Code

Summary by CodeRabbit

• New Features
  • Added support for a new OpenAI-compatible provider with discount pricing.
  • Added gpt-5.5 model availability for the new discount provider.
  • Configured environment variables for the new provider's API key and endpoint configuration.

[chatgpt-codex-connector]

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.

[coderabbitai]

Walkthrough

This PR adds a new openai-discount provider as an OpenAI-compatible endpoint. The changes declare the provider in the model registry, configure required environment variables, add gpt-5.5 model support, and wire the provider through all chat/streaming request and response transformation pipelines to behave identically to the standard openai provider.

Changes

OpenAI Discount Provider Integration

Layer / File(s)	Summary
Provider definition and environment configuration packages/models/src/providers.ts, .env.example	openai-discount provider is declared with environment variables LLM_OPENAI_DISCOUNT_API_KEY and LLM_OPENAI_DISCOUNT_BASE_URL, priority 1.5, and streaming/cancellation enabled.
Model pricing and capability configuration for openai-discount packages/models/src/models/openai.ts	gpt-5.5 model gains an openai-discount provider entry with 0.1 discount factor, matching input/output pricing and feature support (web search, reasoning, JSON schema).
Endpoint and base URL resolution packages/actions/src/get-provider-endpoint.ts	Base-URL loader retrieves openai-discount config from environment; endpoint selector routes openai-discount alongside openai for Responses API and image generation paths.
Request body preparation and parameter handling packages/actions/src/prepare-request-body.ts	openai-discount is added to reasoning-effort native handlers, image generation support, main provider switch, and prompt-cache parameter logic (both Responses API and Chat Completions paths).
Response transformation and streaming format conversion apps/gateway/src/chat/tools/transform-response-to-openai.ts, apps/gateway/src/chat/tools/transform-streaming-to-openai.ts, apps/gateway/src/chat/tools/extract-tool-calls.ts	Response and streaming transforms route openai-discount through OpenAI-compatible handling; tool-call extraction uses same logic as openai.
Streaming token tracking and web-search counting apps/gateway/src/chat/chat.ts	Streaming SSE processing extracts transformedData.usage tokens and counts response.web_search_call.completed events for openai-discount.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Possibly related PRs

• theopenco/llmgateway#2076: Introduced the initial gpt-5.5 model providers that this PR extends with the openai-discount provider entry.
• theopenco/llmgateway#2051: Extended OpenAI-compatible provider routing in chat transformation files (transform-streaming-to-openai.ts, get-provider-endpoint.ts) in a similar pattern to this PR.
• theopenco/llmgateway#2201: Introduced prompt-cache control wiring (prompt_cache_key and prompt_cache_retention) that this PR now extends to openai-discount in prepare-request-body.ts.

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Title check	✅ Passed	The title 'feat(providers): add openai-discount provider' directly and clearly summarizes the main change—adding a new OpenAI-discount provider to the codebase.
Docstring Coverage	✅ Passed	Docstring coverage is 80.00% which is sufficient. The required threshold is 80.00%.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch openai-discount-provider

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[Copilot]

Pull request overview

Adds a new openai-discount provider that mirrors the upstream openai request/response handling but resolves its base URL and API key from environment variables (following the existing anthropic-discount stealth pattern).

Changes:

• New openai-discount provider definition in packages/models with env-driven apiKey/baseUrl, plus an openai-discount mapping (external id gpt-5-5, discount: "0.1", supportsResponsesApi: true) added to the gpt-5.5 model.
• Endpoint resolution (get-provider-endpoint.ts) reads LLM_OPENAI_DISCOUNT_BASE_URL and now keys the Responses-API capability lookup on the actual provider rather than a hardcoded "openai"; request-body transforms in prepare-request-body.ts treat openai-discount like openai on both Responses-API and chat-completions paths.
• Gateway-side OpenAI branches (streaming/non-streaming transform-to-openai, tool-call extraction, usage extraction, web-search counting) extended to also match openai-discount; .env.example documents the two new vars.

Reviewed changes

Copilot reviewed 9 out of 9 changed files in this pull request and generated no comments.

Show a summary per file

File	Description
packages/models/src/providers.ts	Registers openai-discount provider with env-based credentials, priority 1.5, mirroring anthropic-discount.
packages/models/src/models/openai.ts	Adds openai-discount provider mapping for gpt-5.5 (external id gpt-5-5, 10% discount, Responses API).
packages/actions/src/get-provider-endpoint.ts	Resolves base URL from env for openai-discount and uses provider (not hardcoded "openai") when looking up Responses-API capability.
packages/actions/src/prepare-request-body.ts	Treats openai-discount identically to openai for none-effort, image-gen guard, and prompt-cache field handling on both Responses and chat-completions paths.
apps/gateway/src/chat/tools/transform-streaming-to-openai.ts	Routes streaming transform for openai-discount through the OpenAI/Azure branch.
apps/gateway/src/chat/tools/transform-response-to-openai.ts	Routes non-streaming response transform for openai-discount through the OpenAI branch.
apps/gateway/src/chat/tools/extract-tool-calls.ts	Extracts streaming tool-call deltas for openai-discount like openai/azure.
apps/gateway/src/chat/chat.ts	Includes openai-discount in Responses-API usage extraction and web_search_call.completed counting.
.env.example	Documents LLM_OPENAI_DISCOUNT_BASE_URL / LLM_OPENAI_DISCOUNT_API_KEY.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.