feat: Add supportsReasoning property for Z.ai GLM binary thinking mode

[daniel-lxs]

Summary

Implements GLM-4.6 Deep Thinking support for Z.ai as requested in issue #8465.

Changes

New supportsReasoning Property

• Added to ModelInfo schema for binary reasoning models
• Distinct from which handles adjustable reasoning levels

GLM Model Updates

• GLM-4.5 and GLM-4.6: Added (both international and mainland)
• Other GLM models: No reasoning support (per Zhipu AI documentation)

Frontend Integration

• ThinkingBudget component shows simple "Enable reasoning" toggle for supportsReasoning models
• No complex sliders or effort levels for binary reasoning models
• Models without reasoning support show no reasoning controls

Backend Implementation

• ZAiHandler adds to API requests when reasoning is enabled
• Checks both and
• Integrates with existing global reasoning framework

Testing

• All frontend tests pass (11/11)
• All backend tests pass (24/24)
• TypeScript compilation clean
• Comprehensive test coverage for binary reasoning functionality

User Experience

1. Select GLM-4.5 or GLM-4.6: User sees "Enable reasoning" checkbox
2. Toggle reasoning: Simple on/off control activates Z.ai's Deep Thinking mode
3. Other GLM models: No reasoning controls appear

Closes #8465

---

Important

Add supportsReasoningBinary to enable binary reasoning for GLM models, updating ZAiHandler and ThinkingBudget component accordingly.

• Behavior:
  • Add supportsReasoningBinary to modelInfoSchema in model.ts for binary reasoning support.
  • Update internationalZAiModels and mainlandZAiModels in zai.ts to include supportsReasoningBinary for GLM-4.5 and GLM-4.6.
  • ZAiHandler in zai.ts checks supportsReasoningBinary to add thinking parameter in API requests.
• Frontend:
  • ThinkingBudget.tsx shows a simple "Enable reasoning" toggle for models with supportsReasoningBinary.
• Testing:
  • Add tests in zai.spec.ts to verify thinking parameter inclusion based on supportsReasoningBinary.
  • Add tests in ThinkingBudget.spec.tsx to verify UI behavior for binary reasoning models.

^{This description was created by for 8342ce9. You can customize this summary. It will automatically update as commits are pushed.}

[roomote]

Review Summary

I've reviewed the pull request and identified 2 efficiency issues related to duplicate method calls:

Issues to Address

• Refactor createStream method to call getModel() only once and reuse the result
• Refactor completePrompt method to call getModel() only once and reuse the result

---

View Job

[roomote]

Review complete. Found 2 efficiency issues that should be addressed before merging.