LLM Rate Limit & Throughput Calculator
Plan RPM and TPM capacity, traffic growth, utilization and safety headroom using your own account limits.
Enter your actual account/model limits. Initial values are a synthetic example, not provider presets. This is a steady-state average estimate, not a guarantee against throttling.
Ready. Use limits from your own account.
Enable JavaScript to calculate locally.
Find the binding limit.
Average input plus output tokens form the assumed token budget per request. TPM divided by that total gives token-constrained requests per minute; the smaller of that capacity and RPM is effective throughput. Equal capacities are balanced.
Reserve room for variation.
Headroom reduces operational capacity by your chosen percentage. Whole operational requests round down. A traffic multiplier scales expected load; utilization and remaining capacities then describe the growth scenario. Negative remaining values indicate overload. No headroom percentage is universally safe.
Use your real limits.
The starting numbers are synthetic, not provider limits. Account tiers, separate input/output quotas, token reservations, bursts, retries and latency can change actual throughput. This average-load calculator does not predict queues or guarantee freedom from throttling. No account or API connection occurs.