LOCK
Rate limits & quotas
Every Meridian API key is subject to per-minute request and token limits. Limits scale with your plan and reset on a rolling 60-second window.
| Tier | RPM | TPM | Burst | Concurrent |
|---|---|---|---|---|
| Partner / Unlimited | 10,000 | 10,000,000 | 3× baseline for 30 s | 100 |
| Production | 1,000 | 1,000,000 | 2× baseline for 15 s | 25 |
| Trial | 60 | 100,000 | — | 5 |
Handling 429 responses
When you exceed a limit, Meridian returns HTTP 429 Too Many Requests. The response body includes a JSON error object:
{
"error": {
"code": "rate_limit_exceeded",
"message": "RPM limit reached. Retry after 12 s.",
"retry_after_seconds": 12
}
}- →Read the
retry_after_secondsfield and sleep before retrying. - →Implement exponential backoff with jitter to avoid thundering herd.
- →Monitor the
x-ratelimit-remaining-requestsandx-ratelimit-remaining-tokensresponse headers to stay under limits proactively.