Rate Limits
Arena enforces account-level rate limits before dispatching requests to an upstream model provider.
Tiers
Limits are assigned to your account, so all of your API keys share the same bucket. A value of 0 means that dimension is not limited. Tier values are subject to change.
| Tier | RPM | TPM | RPS | TPS |
|---|---|---|---|---|
| Tier 1 | 1,200 | 1,200,000 | 10 | 200,000 |
| Tier 3 | 4,800 | 4,000,000 | 40 | 500,000 |
| Tier 5 | 10,000 | 40,000,000 | 100 | 1,000,000 |
429 responses
When a rate limit is exceeded, Arena returns 429 Too Many Requests without calling the upstream provider.
HTTP/1.1 429 Too Many Requests
Retry-After: 1
X-Arena-Rate-Limit-Dimension: rpm
X-RateLimit-Limit: 1200
X-RateLimit-Remaining: 0
Content-Type: application/json
{
"error": {
"code": "rate_limit_exceeded",
"message": "rate limit exceeded",
"type": "invalid_request_error",
"trace_id": "req_abc123def456"
}
}Use Retry-After for retry timing and the rate-limit headers to understand which configured limit was hit.
Provider capacity
These are Arena account-level limits. They are separate from upstream provider shared-capacity limits, which may also appear in analytics as provider rate-limit signals.
Changing your tier
To request a different rate-limit tier, contact the Arena team through the Support option in the dashboard sidebar.