Skip to main content
Before you start, install the tool and create an API key with prepaid balance in the AsterWise app.
Based on the tool’s documented custom-provider setup. Live AsterWise compatibility has not yet been tested for this tool.

Which API does this use?

This guide selects Chat Completions through api: "openai-completions". The base URL is https://api.asterwise.dev/v1. The client builds the full endpoint: https://api.asterwise.dev/v1/chat/completions. The model name chooses AsterCode or AsterWork, not the protocol. Use the base URL in the configuration below, rather than the full endpoint.

1. Set your API key

2. Add the provider

Create or edit ~/.pi/agent/models.json. Merge this entry into providers if the file already contains other providers:
apiKey is the name of your environment variable, not the key itself. openai-completions selects Chat Completions.

3. Choose your model

Run pi in the same terminal, enter /model, and select asterwise/aster-code. Before your first request, set contextWindow and maxTokens on each model using the app’s Setup recommendations. Pi’s defaults are not AsterWise’s limits.

Check the connection

Ask “Reply with hello” first. Confirm a response appears and find the request in Usage in AsterWise. Then try a small coding task before a larger project. For authentication errors, check the key and restart the tool from the terminal where you exported it. See troubleshooting.

Official reference

Configuration reviewed on September 16, 2026. Pi custom models.