fix(models): add glm 4.5v and adapt pricing

[steebchen]

Revised input/output pricing, reasoning, and tools support for ZAI models to reflect updated specifications. Models impacted include glm-4.5, glm-4.5-air, glm-4.5-x, glm-4.5-airx, glm-4.5-flash, and glm-4-32b-0414-128k.

Summary by CodeRabbit

• New Features
  • Added glm-4.5v model with vision, reasoning, tool use, streaming, and large context support.
• Changes
  • Updated pricing across glm-4.5 family (glm-4.5, 4.5-air, 4.5-x, 4.5-airx, 4.5-flash) and glm-4-32b-0414-128k.
  • Enabled tool use across these models; reasoning enabled on select variants.
• Tests
  • Minor comment added in an e2e test; no behavioral impact.

[coderabbitai]

Walkthrough

Adds a comment in an e2e test regarding empty text handling. Updates Z AI model registry: introduces glm-4.5v and revises pricing and capability flags (reasoning/tools/vision/streaming/context) for multiple glm-4.5* providers in zai.ts. No exported API shape changes beyond the new model entry.

Changes

Cohort / File(s)	Summary
E2E tests apps/gateway/src/api.e2e.ts	Inserted a comment after an empty text field in a test message; no logic, data, or flow changes.
Z AI models packages/models/src/models/zai.ts	Added model glm-4.5v (vision, reasoning, tools, streaming, context 128k, pricing set). Updated glm-4.5, -air, -x, -airx, -flash, and glm-4-32b-0414-128k pricing and capability flags (reasoning/tools as specified); no other structural changes.

Sequence Diagram(s)

sequenceDiagram
  autonumber
  actor Client
  participant Gateway
  participant ZAI as Z AI Provider

  Client->>Gateway: Request with model=glm-4.5v | glm-4.5* …
  alt Model supports tools/reasoning/vision
    Note over Gateway: Set flags per model (tools/reasoning/vision/streaming)
    Gateway->>ZAI: Forward request with capabilities + pricing metadata
    ZAI-->>Gateway: Response (streaming if enabled)
  else Model without reasoning
    Gateway->>ZAI: Forward request (tools as applicable)
    ZAI-->>Gateway: Response
  end
  Gateway-->>Client: Deliver response

Loading

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~10 minutes

Possibly related PRs

• feat(models): add Z AI models and provider integration #566 — Prior updates to Z AI model/provider definitions that this PR extends (adds glm-4.5v and revises provider fields/pricing).
• test(api): add e2e tests for tool call models #580 — Touches the same files and model "tools" capability used by e2e tool-call tests, aligning with this PR’s capability updates.

Suggested reviewers

• smakosh

✨ Finishing Touches

• 📝 Generate Docstrings

🧪 Generate unit tests

• Create PR with unit tests
• Post copyable unit tests in a comment
• Commit unit tests in branch feat/zai

---

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.}

[coderabbitai]

Actionable comments posted: 0

🧹 Nitpick comments (2)

apps/gateway/src/api.e2e.ts (1)

1274-1276: Avoid empty text parts in multi-part content to reduce flakiness.

Some providers reject empty text items in content arrays. If the goal isn't to test this edge case, prefer a single space or drop the segment.

Apply this minimal tweak if you see intermittent 4xx:

-            text: "", // empty text – note this may need special handling
+            text: " ", // intentionally blank but non-empty to avoid provider validation failures

packages/models/src/models/zai.ts (1)

130-142: Zero-priced glm-4.5-flash — decide whether it should be treated as “free”.

Test filtering uses the model-level free flag, not prices. If you intend Flash to be excluded in non-full runs as a free model, mark it accordingly.

Optional change:

   {
     id: "glm-4.5-flash",
     name: "GLM-4.5 Flash",
     family: "glm",
     deprecatedAt: undefined,
     deactivatedAt: undefined,
+    free: true,
     providers: [

📜 Review details

Configuration used: CodeRabbit UI

Review profile: CHILL

Plan: Pro

📥 Commits

Reviewing files that changed from the base of the PR and between 64d7366 and 7ef329e.

📒 Files selected for processing (2)

• apps/gateway/src/api.e2e.ts (1 hunks)
• packages/models/src/models/zai.ts (6 hunks)

🧰 Additional context used

📓 Path-based instructions (7)

**/*.{js,jsx,ts,tsx}

📄 CodeRabbit inference engine (.github/copilot-instructions.md)

Use localStorage instead of cookies for client-side data persistence

Files:

• apps/gateway/src/api.e2e.ts
• packages/models/src/models/zai.ts

**/*.{js,ts}

📄 CodeRabbit inference engine (.github/copilot-instructions.md)

**/*.{js,ts}: Use drizzle with the latest object syntax for database operations
For read queries, always use db().query.<table>.findMany() or db().query.<table>.findFirst()

Files:

• apps/gateway/src/api.e2e.ts
• packages/models/src/models/zai.ts

**/*.{ts,tsx}

📄 CodeRabbit inference engine (.cursor/rules/general.mdc)

Never use as any or : any in TypeScript files.

Files:

• apps/gateway/src/api.e2e.ts
• packages/models/src/models/zai.ts

apps/{api,gateway}/**/*.{ts,tsx}

📄 CodeRabbit inference engine (AGENTS.md)

apps/{api,gateway}/**/*.{ts,tsx}: Use Drizzle with the latest object syntax for database operations
For read queries, use db().query.

.findMany() or db().query..findFirst()

apps/{api,gateway}/**/*.{ts,tsx}: Use the Hono framework for backend HTTP services in apps/api and apps/gateway
Use Zod for request/response validation in backend routes and handlers
Maintain OpenAPI/Swagger documentation for backend APIs

Files:

• apps/gateway/src/api.e2e.ts

apps/gateway/**/*.{ts,tsx}

📄 CodeRabbit inference engine (AGENTS.md)

In apps/gateway (Hono), always use Hono + Zod + OpenAPI for validation and typesafety

Files:

• apps/gateway/src/api.e2e.ts

**/*.e2e.ts

📄 CodeRabbit inference engine (CLAUDE.md)

Name end-to-end tests with the .e2e.ts suffix (run by pnpm test:e2e)

Files:

• apps/gateway/src/api.e2e.ts

**/*.{spec,e2e}.ts

📄 CodeRabbit inference engine (CLAUDE.md)

Use Vitest as the test framework for unit and E2E tests

Files:

• apps/gateway/src/api.e2e.ts

⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (2)

• GitHub Check: e2e / run
• GitHub Check: build / run

🔇 Additional comments (3)

packages/models/src/models/zai.ts (3)

14-23: glm-4.5 pricing/capabilities updated — verify reasoning output semantics.

Numbers/flags look consistent. If glm-4.5 does not emit explicit reasoning output, set reasoningOutput: "omit" like 4.5v to keep e2e assertions aligned.

---

28-50: New glm-4.5v (vision + reasoningOutput: "omit") — LGTM.

Configuration matches tests that gate reasoning output on reasoningOutput !== "omit" and include vision inputs.

---

61-70: Pricing/tools updates across glm-4.5-air/x/airx and glm-4-32b — looks good.

Flags (tools: true, reasoning on the right variants) align with test filters; no type issues with satisfies.

Also applies to: 84-93, 107-116, 153-162