Error codes

Code	Description
400	Invalid request — e.g. prompt + max_tokens exceeds the model's context window
401	Invalid or missing API key
429	Token limit exceeded or all slots busy
503	Model temporarily unavailable

A 429 response includes details about the window and when quota is reclaimed, plus a Retry-After header:

json

{
  "detail": {
    "message": "Daily token limit exceeded",
    "window": "daily",
    "tokens_used": 250000,
    "tokens_limit": 250000,
    "retry_after_seconds": 3600,
    "reset_at": "2026-08-30T00:00:00+00:00"
  }
}

Handle 429 and 503 with backoff

Both are transient. Respect the Retry-After header (or retry_after_seconds) and retry — for 429, quota is continuously reclaimed (rolling hourly windows) or resets at midnight UTC (daily limits), see Plans and quotas.

Rate limit headers

Every response includes headers showing your current token usage:

Header	Description
X-RateLimit-Limit-Tokens	Token limit for the period
X-RateLimit-Used-Tokens	Tokens used
X-RateLimit-Remaining-Tokens	Tokens remaining

Token warnings

At 80% of the quota a warning field is included in the API response
At 90% a warning email is sent (max one per day)

Track your consumption in real time via GET /v1/usage or in the dashboard.

Error codes and rate limits

Error codes

Rate limit headers

Token warnings