API reference
Base URL: https://graze-api-production.up.railway.app. Authenticate with Authorization: Bearer <your Graze key>.
POST /v1/chat/completions
OpenAI chat completions format, including streaming ("stream": true). model is graze/auto or any model in How routing works. If you omit max_tokens, Graze uses 1,024.
GET /v1/models
The models Graze serves. No key needed.
GET /v1/balance
Your agent's name and balance in USD.
GET /stats
Public totals behind the live dashboard: calls, tokens, spent, saved, and recent calls. No key needed.
GET /api/token-config
The official $GRAZE addresses per chain and the launch time. Never cached. No key needed.
{"chains": {"solana": null, "robinhood": "0x..."}, "launch_at": "2026-09-26T15:00:00Z", "symbol": "GRAZE"}
Errors
| Status | Meaning |
|---|---|
400 |
Bad request: unknown model, invalid max_tokens |
401 |
Missing or invalid Graze key |
402 |
Balance too low for this call's hold |
5xx |
Every model failed; the provider's error is passed through, and nothing is charged |