Cherry Studio

Configure a local Cherry Studio provider with the correct OpenAI base and model list.

Install and prepare

Get the Windows, macOS or Linux installer from Cherry Studio releases, install it and launch the desktop app. Create a key at routerra.ai/keys for this client and copy an allowed model ID from your gateway model list. No Docker or local inference server is needed; this setup keeps the key in the client’s provider settings.

Configure the connection

Open Settings → Model Services → + Add Provider. Name it routerra.ai, choose OpenAI for the endpoint, set its base URL to https://routerra.ai/v1, and enter YOUR_API_KEY. In versions showing API Address, use https://routerra.ai/v1; inspect the request URL preview to ensure it resolves to /v1/chat/completions. Newer versions expose separate protocol endpoints: configure the OpenAI chat endpoint, not Anthropic or Responses. Save and enable the provider.

OpenAI base URL: https://routerra.ai/v1
API Key: YOUR_API_KEY
Model ID: YOUR_MODEL_ID
Generated chat URL: https://routerra.ai/v1/chat/completions

Select a model and resolve errors

Click Get model list and add an allowed model, or use + to add YOUR_MODEL_ID manually. Select that provider/model in a new chat and send a short text. For 404 errors, check the final URL for a duplicated /v1 or a missing /chat/completions path; for an empty list, check the provider is enabled and add the exact ID manually. A successful key check does not guarantee access to every model.