v.0.8.1-beta

v0.8.1-beta

New features added to extended thinking and streaming.

Bug Fixes

• Fixed model constraints not being applied for Anthropic (undefined variable)
• Fixed per-action reasoning config not being used when set in action editor
• Fixed model capability matching for versioned model names (e.g., claude-sonnet-4-5)
• Fixed reasoning indicator appearing in debug mode when setting was disabled
• Fixed Gemini thinking API format (now uses correct camelCase REST format)

Improvements

• Simplified streaming reasoning display for better e-ink compatibility
• OpenAI reasoning models now show indicator with effort level (e.g., "Reasoning requested (medium)")
• Debug info stored per-message - shows config that was actually used, not current settings

Code Quality

• Removed duplicate request body building code in API handlers
• Centralized reasoning default values
• Consolidated duplicate UI helper functions

Full Changelog: v0.8.0...v.0.8.1-beta