CAMEL changelog: what's new each month
Every stable CAMEL release summarised by month: the highlights, new features, improvements, fixes and anything you need to act on. 3 months covered; the current month updates daily.
March 2026
1 release: 0.2.90
March brought expanded model support and improved thinking capabilities across multiple AI providers. The release focused on compatibility enhancements and better error handling.
Highlights
- Interleaved thinking now available for Kimi K2 and GLM models
- Added support for Minimax M2.5 and GLM5 models
- Improved error messaging when ffmpeg is missing
- Enhanced capture of rich output from code execution results
New
- Interleaved thinking support for Kimi K2
- Interleaved thinking support for GLM
- Minimax M2.5 model support
- GLM5 model support
Improved
- Runtime error now thrown when ffmpeg binary is absent, instead of failing silently
- Rich output from e2b execution results is now properly captured
- Workforce quality check threshold adjusted to 60
Fixed
- Typo in documentation corrected ('seperated' to 'separated')
- Pre-commit configuration error resolved
- Rich output capture from code execution results
February 2026
3 releases: 0.2.86 → 0.2.89
February brought improvements to browser automation capabilities, expanded model support, and new agent skills. The month focused on enhancing detection and caching features while unifying runtime cleanup across the platform.
Highlights
- Added prompt caching support for improved performance
- Enhanced browser toolkit with better page loading detection
- Introduced browser visual pixel mode for more accurate interactions
- Added support for Kimi K2.5 model
- New ChatAgent skills for expanded agent capabilities
New
- Prompt caching support
- Browser visual pixel mode
- Kimi K2.5 model support
- ChatAgent skills
Improved
- HybridBrowserToolkit page loading detection enhanced
- Unified cleanup process across all runtimes
- Quality check prompt improvements
Fixed
- Agent tests updated to remove deprecated Claude models
January 2026
3 releases: 0.2.83 → 0.2.85
January brought improvements to agent reliability and expanded integrations, with fixes to token tracking, memory handling, and concurrent execution. New capabilities for email and function calling were added alongside enhancements to browser automation and response handling.
Highlights
- Added Microsoft Outlook mail actions for sending and managing emails.
- Fixed token usage tracking in streaming mode.
- Improved browser automation with absolute coordinate positioning replacing relative navigation.
- New Gemma function integration for enhanced function calling.
- Fixed ChatAgent memory issues and agent pool race conditions.
New
- Microsoft Outlook integration for mail actions.
- Gemma function integration support.
Improved
- Browser automation now uses absolute coordinate positioning instead of relative navigation.
- Long toolkit responses are truncated to improve usability.
- Browser popup handling improved.
- Model documentation updated with Zhipu model support.
Fixed
- Token usage tracking in streaming mode.
- ChatAgent memory setter issue.
- Sliding window message handling when window size is zero.
- LLM guard functionality.
- Agent pool race condition.
- Daytona interface compatibility.
Summaries are written automatically from the official release notes (full changelog ↗); check the original notes before relying on a detail. CAMEL: pricing, features and alternatives · All changelogs