Trae

ByteDance's AI IDE you can point at TokenMix

What it is

Trae is ByteDance's AI-native IDE, built as an AI-first code editor with chat-based generation, an agent builder, and a SOLO mode that acts as an AI engineer. It started as MarsCode and is now Trae, available as a standalone desktop IDE and as a plugin for VS Code and JetBrains. The standalone IDE lets you add your own custom model, which is how you connect it to TokenMix.

Before you start

Download and install the Trae IDE from its official site (trae.ai). Create a TokenMix account, open the console, and generate an API key (it looks like sk-tm-...). Pick a coding model to use, for example deepseek-v4-pro, qwen3-coder-plus, or glm-5.1.

Connecting Trae to TokenMix

In the Trae IDE, open the model picker — click the model name in the chat box, or go to the top-right avatar, then AI Feature Management, then Models — and choose Add Model. In the dialog, set the provider to OpenAI, turn the Base URL "Full URL" toggle on and enter https://api.tokenmix.ai/v1/chat/completions (or leave the toggle off and enter https://api.tokenmix.ai/v1), enter your sk-tm-... key for the API key, and put the model you want in the Model ID field, for example deepseek-v4-pro. Save the model.

Verify it works

Select your new model in Trae's model picker and ask it to explain or write a small piece of code. If it responds, Trae is now calling your TokenMix model.

FAQ

Trae's custom model uses the OpenAI chat-completions format, which is what TokenMix's endpoint speaks, so any chat model TokenMix serves works — add more models by repeating Add Model with a different Model ID. Your key is stored locally. If the model fails to connect, check the Full URL toggle: when it is on, the URL must end in /v1/chat/completions; when off, use the /v1 root.