Skip to main content

Roo Code 3.25.11 Release Notes (2025-08-11)

This release enhances GPT-5 support, adds the new IO Intelligence provider, and includes several new features, quality-of-life improvements, and bug fixes.

Enhanced GPT-5 Support​

We've enhanced our GPT-5 integration, enabling you to leverage more advanced capabilities for streaming, multi-turn conversations, and efficient token management. This release also adds support for the new GPT-5 models from OpenAI, including gpt-5-2025-08-07, gpt-5-mini-2025-08-07, and gpt-5-nano-2025-08-07.

📚 Documentation: See the OpenAI Provider documentation for more details.

New IO Intelligence Provider​

We've added IO Intelligence as a new provider, giving you access to a wide range of AI models like Llama, DeepSeek, Qwen, and Mistral through a unified API (#6875).

📚 Documentation: See the IO Intelligence Provider documentation for more information.

Codex Mini Model Support​

We've added support for the codex-mini-latest model in the OpenAI Native provider, allowing you to leverage its specialized code-related capabilities directly (thanks KJ7LNW!) (#6931).

QOL Improvements​

  • codebase_search Tool: Clarified that the path parameter is optional and the tool searches the entire workspace by default (#6877).
  • @roo-code/cloud Linking: Improved the developer workflow by allowing @roo-code/cloud to be directly linked from a local repository, with HMR support (#6799).
  • Chat Input Focus: The chat input is now automatically focused when creating a new chat from the extension's top menu (#6689).
  • Token Usage Reporting: Fixed an issue where token usage and cost were underreported, providing more accurate cost tracking (thanks chrarnoldus!) (#6122).

Bug Fixes​

  • MCP Server Startup: Fixed an issue where MCP servers would fail to start and removed unnecessary refresh notifications (#6878).
  • Tool Repetition Detector: Fixed a bug where setting the "Errors and Repetition Limit" to 1 would incorrectly block the first tool call (thanks NaccOll!) (#6836).
  • MCP Error Messages: Fixed an issue where MCP server error messages were displaying raw translation keys (#6821).
  • apply_diff Tool: Fixed a bug that caused XML parsing errors when working with complex XML content (#6811).
  • max_tokens Calculation: Fixed an error for models with very large context windows where requests would fail due to incorrect calculation of maximum output tokens (thanks markp018!) (#6808).
  • Scroll Jitter: Fixed a scroll jitter issue that occurred during message streaming, especially with code blocks (#6780).
  • MCP Server Refresh: Fixed an issue where MCP servers would unnecessarily refresh when unrelated settings were saved (#6779).
  • AWS Bedrock Connection: Fixed a connection issue when using AWS Bedrock with LiteLLM (#6778).

Provider Updates​

  • Fireworks: Added support for four new models: GLM-4.5, GLM-4.5-Air, gpt-oss-20b, and gpt-oss-120b (thanks alexfarlander!) (#6784).