Skip to content
Fast Inference
DOCS / Grok
Integration

Grok

Track the planned Grok adapter and use Fast Inference's OpenAI-compatible API directly in the meantime.

Grok is part of the fast-web integration catalog, but the released fast CLI does not currently include a Grok adapter.

Use the API directly

You can use the same models and project key from any OpenAI-compatible client:

curl
curl https://api.inference.net/v1/chat/completions \
  -H "Authorization: Bearer $INFERENCE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "glm-5.2",
    "messages": [{ "role": "user", "content": "Hello" }]
  }'

Create the key in API keys, and copy a callable model ID from the model catalog. Requests made with that key appear in Usage under the same first project.

What release requires

The integration will move out of preview when the adapter can:

  • configure Grok through a stable, supported contract;
  • keep credentials out of shell history;
  • snapshot and restore the original configuration;
  • report the selected model and config path through fast grok status; and
  • complete an end-to-end request against the selected fast-web project.

Available today

Use Claude Code, Codex, OpenCode, or Pi for a released one-command integration.