[ENHANCEMENT] Add support for GLM-5.1 via Z.ai provider

[akfrolov]

Problem (one or two sentences)

Roo Code has no support of newly released GLM-5.1 via GLM Coding Plan.

Context (who is affected and when)

All users who have Z.ai API and Coding plan subscriptions are not able to use glm-5.1 model in Roo.

Desired behavior (conceptual, not technical)

GLM-5.1 should be available among models in Z.ai provider.

Constraints / preferences (optional)

No response

Request checklist

• I've searched existing Issues and Discussions for duplicates
• This describes a specific problem with clear context and impact

Roo Code Task Links (optional)

No response

Acceptance criteria (optional)

No response

Proposed approach (optional)

No response

Trade-offs / risks (optional)

No response

I've analyzed this issue and here's my proposed implementation plan:

The codebase already has GLM-5 defined in the Z.ai provider. Adding GLM-5.1 follows the same pattern. The changes would be:

1. packages/types/src/providers/zai.ts - Add "glm-5.1" entry to both internationalZAiModels and mainlandZAiModels objects, modeled after GLM-5 (with thinking/reasoning support, ~202k context window). Pricing and exact specs would need to match the official Z.ai/BigModel docs for GLM-5.1.

2. src/api/providers/__tests__/zai.spec.ts - Add test cases for the new GLM-5.1 model in both international and China API line test suites, verifying model info is returned correctly (similar to existing GLM-4.7 and GLM-5 tests).

Questions before proceeding:

• Do you have the official GLM-5.1 pricing and context window specs? If not, I'll use the same values as GLM-5 as a starting point (202,752 context window, 16,384 max output tokens, same pricing tiers).
• Should GLM-5.1 support thinking/reasoning like GLM-5 does (with supportsReasoningEffort: ["disable", "medium"])?
• Does GLM-5.1 support vision (images), or is it text-only like GLM-5?

Please mention @roomote in a comment saying that you approve this plan, and I'll proceed with the implementation.

[akfrolov]

It should be exactly as GLM-5. However, API is unavailable for now, GLM-5.1 is only active for GLM Coding Plan.

[zeropointnine]

bump

[alefteris]

@roomote Update your plan using the information bellow:

• Official GLM-5.1 release blog post: https://z.ai/blog/glm-5.1 .
• Model guide docs: https://docs.z.ai/guides/llm/glm-5.1 .
• Updated pricing page with the new model: https://docs.z.ai/guides/overview/pricing . For GLM-5.1 the pricing is:
  • Input: $1.4
  • Cached Input: $0.26
  • Output: $4.4
• GLM-5.1 Context window: 200K
• GLM-5.1 Max output tokens: 128K
• GLM-5.1 uses the same thinking/reasoning options that GLM-5 does
• GLM-5.1 supports only text (like GLM-5)

[alefteris]

Related Pull Request: #12036

I've seen your comment with the GLM-5.1 specs and will work on adding the model to the Z.ai provider. I'll open a PR to share progress shortly.

Opened a draft PR with my attempt at adding GLM-5.1 support: #12077

The PR adds glm-5.1 to both international and mainland model lists with the specs from your comment (200K context, 128K max output, thinking support, text-only). Mainland pricing is estimated from existing model ratios since only international pricing was provided. Feedback welcome!

[kommandantredundant]

any news on this? is anyone able to finish this up? I'd love to run glm-5.1 in the vscode extension but I can only see glm-5. 5.1 is available to all coding plans right now

[akfrolov]

FYI, you can just type glm-5.1 in model selector, then press "Use custom: glm-5.1" (or something similar) and Roo will use this new model.

[karpo518]

FYI, you can just type glm-5.1 in model selector, then press "Use custom: glm-5.1" (or something similar) and Roo will use this new model.

I'm not sure this works correctly. When I specify the glm-5.1 model, the agent tells me I'm talking to the glm-4.6 model. I have no way to verify which model to send the request to.