Changelog

Follow up on the latest improvements and updates.

RSS

  1. Optimize the automatic update process and improve the experience
  2. Fixed an issue where automatic updates could cause downgrades
  3. Fix model visual ability detection anomalies
  4. Fixed an issue where a new session was not automatically selected after creating a session from a partner
  5. When a message is generated, the submit button automatically switches to a stop button
  6. Improve the way API error messages are displayed
  • Supports Gemini multi-modal output
  • Supports more message context restriction options
  • Thoughts are automatically folded
  • Check for updates regularly, not at startup
  • Fix some model tool call issues
  • Fixed an issue with understanding ollama images
  • Fixed an issue where XAI only supports the Grok-Beta model; other models can now be used normally
  • Fixed an issue where some models did not show the thought process during inference
  • Fixed a bug where the app did not close properly under certain circumstances
  • Optimize the online search experience on mobile devices to improve stability and fluency
  • Other performance optimizations and bug fixes
  • Refactored settings UI
  • Fixed some LLM API call issues
  • Fixed issues with model temperature and ToPP parameters