1.13.0a7
The latest pre-release enables multimodal vision for OpenAI's newest models and introduces a major UI framework.
CrewAI Inc has pushed a significant pre-release update, version 1.13.0a7, to its flagship open-source framework for orchestrating role-playing, autonomous AI agents. The update is headlined by two major additions: formal support for OpenAI's newest model families and the integration of a new UI extension framework. This release specifically addresses the growing need for agents that can process and understand visual information alongside text.
Technically, the update fixes 'multimodal vision prefixes' by adding explicit support for GPT-5 and the o-series (like GPT-4o). This means developers can now reliably use these advanced models for tasks requiring image analysis within their CrewAI agent crews. Furthermore, the introduction of the A2UI extension (with support for versions 0.8 and 0.9) provides a structured way to build user interfaces for agents, complete with schemas and documentation, moving the framework beyond pure backend orchestration.
The release also includes standard bug fixes and documentation updates, maintaining the project's rapid development pace. With over 47.8k stars on GitHub, CrewAI is a central tool for developers building complex, collaborative AI systems. This update ensures those systems can leverage the very latest model capabilities from leading providers, keeping the framework at the forefront of the AI agent ecosystem.
- Adds official multimodal vision support for OpenAI's GPT-5 and o-series models, fixing previous integration issues.
- Introduces the new A2UI extension with v0.8/v0.9 support, schemas, and docs for building agent interfaces.
- Maintains CrewAI's position as a leading open-source framework, now with 47.8k GitHub stars, for orchestrating AI agents.
Why It Matters
Enables developers to build next-gen AI agents that seamlessly combine advanced visual reasoning with autonomous task execution.