LibreChat

Open-source, self-hosted chat UI for every AI provider — connect your team to TokenMix.

What is LibreChat

LibreChat is an open-source, self-hosted chat UI that brings every AI provider into one polished interface — conversations, presets, prompts, file uploads, and multi-user support. Point it at TokenMix through a custom endpoint and your whole team can use the full model catalog.

Before you start

• A running LibreChat instance (self-hosted via Docker or source)
• A TokenMix API key (sk-tm-...) from the console

Add TokenMix as a custom endpoint

1. Open your librechat.yaml config file (in the project root).
2. Under endpoints.custom, add TokenMix:

endpoints:
  custom:
    - name: "TokenMix"
      apiKey: "${TOKENMIX_API_KEY}"
      baseURL: "https://api.tokenmix.ai/v1"
      models:
        default: ["deepseek-v4-flash", "claude-sonnet-4.6", "gpt-5.5"]
        fetch: true
      titleConvo: true
      titleModel: "deepseek-v4-flash"

3. Add the key to your .env: TOKENMIX_API_KEY=sk-tm-your-key
4. Restart LibreChat.

Start chatting

Open LibreChat, pick TokenMix from the endpoint menu, choose a model, and send a message.

Verify

If the model replies, you are connected. Track usage in your TokenMix console.

Common questions

Does baseURL need /v1? Yes — use https://api.tokenmix.ai/v1. LibreChat appends /chat/completions automatically.

Can each user bring their own key? Yes. Set apiKey: "user_provided" instead of an env variable, and LibreChat will ask each user for their own TokenMix key.

Which models can I use? Any in the TokenMix catalog. List them in models.default, or set fetch: true to load them automatically.