Clients

Claude Desktop

Connect Claude Desktop to GreenTokens through third-party inference.

Set up

Claude Desktop can send its requests to an LLM gateway that speaks the Anthropic Messages API, which GreenTokens does. Configure it from inside the app:

Enable developer mode: Help → Troubleshooting → Enable Developer Mode. On Windows, open the application menu (☰) in the top-left of the sign-in screen first.
Open Developer → Configure Third-Party Inference…
In Connection, set Inference provider to Gateway and fill in the gateway credentials:
- Gateway base URL: https://api.greentokens.io
- Gateway API key: your sk-gt-… key
- Credential kind: Static API key
- Gateway auth scheme: Bearer (the default)
Click Apply Changes. Claude Desktop relaunches using GreenTokens.

Enter the base URL exactly as shown: no /v1 and no trailing slash. Claude Desktop adds /v1/messages itself, and requires https.

Models

Claude Desktop reads the model list from GreenTokens automatically and shows the Claude models in its picker. To show only specific models instead, list their IDs in the Models section of the same window:

Models (optional)

claude-opus-5-5
claude-sonnet-5
claude-haiku-4-5

Keeping costs in check

Give Claude Desktop its own API key with a daily spend limit, so long Cowork or Code sessions stay within a budget.
Claude Desktop reuses the system prompt and earlier turns between requests, which GreenTokens bills at the lower cached-input price.