You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
refactor(model-caps): enhance model capabilities resolution and validation
- Remove xAI chat model configs from YAML (now server-driven)
- Add supports_strict_tools and supports_max_completion_tokens fields
- Implement model capability validation with reasonable limits clamping
- Improve canonical model name parsing for multi-segment paths (OpenRouter, Gemini)
- Add refresh lock to prevent concurrent model caps updates
- Make cache I/O async with proper directory creation
- Update OpenAI/Gemini model capabilities (strict tools, no temperature)
- Add comprehensive tests for resolution, validation, and patterns
- Fix default model validation to use full model IDs
Fixes #model-capabilities-resolution
0 commit comments