Limits

Rate limits protect platform stability and ensure fair usage.

Unless otherwise agreed, limits are enforced per account linked to the API key.

Endpoint Rate Limits

Endpoint	Sustained limit	Burst limit
/check	100 requests/minute	5 requests/second
/status	5 requests/minute	1 request/second
/wallet	5 requests/minute	1 request/second

Both limits apply: a request can be rejected if either the per-minute or per-second threshold is exceeded.

When You Hit a Limit

If your client exceeds a limit, the API returns a rate-limit error (typically HTTP 429).

Recommended client behavior:

• Retry with exponential backoff and jitter.
• Reduce concurrency and request bursts.
• Respect Retry-After when present.

Word Balance and Quota

Use /wallet to inspect the number of words still available for the authenticated user. This helps you display remaining quota, pause or defer /check requests when the balance is low, and avoid unexpected failures on exhausted usage plans.

Implementation Guidance

• Implement client-side throttling for both per-second and per-minute windows.
• Keep /status polling minimal; avoid aggressive health-check loops.
• Queue /check traffic and drain at a controlled rate.
• Monitor rate-limit responses and alert on repeated throttling.
• Call /wallet sparingly for pre-flight checks rather than in tight polling loops.

Need Higher Throughput?

Contact support at crew@languagecheck.ai with your expected request profile (average RPM, peak RPS, and use case).