Continue

Moth DocsTutorials

Continue reads its models from one file, where Moth goes in as an OpenAI provider with its own base URL.

Open ~/.continue/config.yaml, or on Windows %USERPROFILE%\.continue\config.yaml, from the gear beside Local Config in Continue's panel, and add the models below.

config.yaml
name: Moth
version: 0.0.1
schema: v1

models:
  - name: Claude Sonnet 4.6 (Moth)
    provider: openai
    model: claude-sonnet-4.6
    apiBase: https://api.runmoth.si/v1
    apiKey: moth_your_key_here
    roles:
      - chat
      - edit
      - apply
    capabilities:
      - tool_use
    defaultCompletionOptions:
      contextLength: 200000
      maxTokens: 8192
  - name: Gemini 3.8 Flash (Moth)
    provider: openai
    model: gemini-3.8-flash
    apiBase: https://api.runmoth.si/v1
    apiKey: moth_your_key_here
    capabilities:
      - tool_use
    defaultCompletionOptions:
      contextLength: 1048576
      maxTokens: 8192
  • capabilities with tool_use lets Agent mode use tools with a model Continue doesn't know.
  • contextLength is the model's context_length from the catalog.
  • Don't give a Moth model the embed or rerank role, since Moth doesn't serve embeddings.
  • On a GPT model add useResponsesApi: false, since Continue sends some GPT names to OpenAI's Responses API, which Moth doesn't serve.
  • To keep the key out of the file, write ${{ secrets.MOTH_API_KEY }} in its place.

To check it worked, pick the model in Continue's chat and send a message, then switch to Agent mode and ask it to list the files in your project.

A request a model can't take is refused with a 400 that says why. Some tools send stop sequences, which the GPT, Grok, DeepSeek and Kimi models refuse, so pick a Claude or Gemini model if you see that. What is removed lists every setting and what happens to it.