Skip to main content

Continue

Point Continue at the gateway as an OpenAI-compatible provider. A live Continue process has not been run against the gateway (not verified in the compatibility matrix).

Config​

Chat uses the chat completions endpoint. Autocomplete is a separate model on the completions endpoint:

{
  "models": [
    {
      "title": "Yabanin chat",
      "provider": "openai",
      "model": "gpt-4o",
      "apiBase": "https://gateway.example.internal/v1",
      "apiKey": "yb_…"
    }
  ],
  "tabAutocompleteModel": {
    "title": "Yabanin complete",
    "provider": "openai",
    "model": "gpt-4o",
    "apiBase": "https://gateway.example.internal/v1",
    "apiKey": "yb_…"
  }
}

Set the autocomplete model so Continue calls POST /v1/completions (prompt and suffix). Chat calls POST /v1/chat/completions.

Bring your own provider key (BYOK)​

Your org admin adds the upstream provider key once in the admin console (Settings → Provider keys — see BYOK). When the model routes through the org credential, the client needs no changes: keep the yb_… gateway key above and the gateway authenticates upstream with the org key.

Limits​

  • The completions row in the matrix is fill-in-the-middle only, not chat.
  • Fixture coverage is not a live Continue run.

Check​

curl -sS -o /dev/null -w '%{http_code}\n' \
  -H "Authorization: Bearer yb_…" \
  https://gateway.example.internal/v1/models

Replace yb_… with the real key. 200 checks the gateway, not Continue.