TRAE is an AI development environment that combines agent-driven task execution with a conventional code editor. It ships as TRAE Desktop (macOS 12.0+ and Windows 10/11), TRAE Web and TRAE Mobile, and projects, tasks and settings stay in sync across them. TRAE says TraeCode and TraeWork have been merged into this single product.
TRAE Desktop has two windows. The Agent window manages multiple agent tasks across projects and runtime environments and can hand long-running tasks to the cloud. The IDE window is a code editor with VS Code extension support, debugging, Git source control, AI code review and Remote SSH or WSL development. It offers IDE mode for hands-on work and SOLO mode, in which the AI plans a task and moves through code generation, testing, preview and change summaries.
Built-in models include OpenAI's GPT-6 and GPT-5.x families, Google's Gemini 3.1 Pro and Gemini 3 Flash previews, Seed-2.1-Turbo, GLM-5.2, DeepSeek-V4-Flash, and Kimi and MiniMax models. The Seed, MiniMax and GLM series are not available to users in the United States. Custom models can be added through preset providers or any endpoint that speaks the OpenAI Chat Completions or Anthropic Messages format, for example Claude models. Custom agents can be configured with prompts, MCP servers and toolsets, and CUE provides completion, multi-line edits and next-edit prediction.
Privacy mode stops TRAE from using chats, code snippets and AI outputs for analysis, product improvement or model training, and codebase files stay on the local device. Agent-generated commands can run in a sandbox with file-access controls and high-risk command blocking. Paid plans meter usage in dollars at TRAE's published per-token model prices.




