Byte AI Engineer "Trae" Releases New Version

Yesterday, Trae, the AI engineer of ByteDance, released a new version, updating a number of features: support for intelligent collaboration mode: summon the needed Agent and give contextual content, AI can actively understand and quickly respond; upgrade of the intelligent body ecology: Trae creates an open intelligent body ecology, and in the face of different demand scenarios, the user is free to configure the intelligent body system; and contextual comprehension ability is upgraded: Trae has upgraded its comprehension ability, supporting the comprehension of information within the IDE such as code repository and external information such as network search and uploading documents, from which the AI will automatically put together a complete demand puzzle through contextualized content; Trae IDE supports the integration of various external tools, such as the call of external resources through the MCP. It is understood that Trae officially launched the domestic version in March, which is configured with Doubao-1.5-pro model and supports switching the full-blooded version of DeepSeek R1 and V3 models. According to the official introduction, Trae is positioned as an "intelligent collaborative AI IDE", with the core concept of "human-machine collaboration and mutual enhancement", and it is very adaptable to the whole scenario of the development process, such as code completion, code comprehension, Bug fixing, and code generation based on natural language.