Anthropic · Rate Limits

Anthropic Rate Limits

Reconciled rate limits for the Anthropic Messages, Batches, and Managed Agents APIs. Token-bucket algorithm; only uncached input tokens count toward ITPM on most models.

12 Limits Throttle: 429 Quota: 429
AIRate LimitingQuotas

Limits

Policies

Cache-aware ITPM
On most models, cache_read_input_tokens do NOT count toward ITPM, making prompt caching an effective way to increase throughput.
Auto tier advancement
Tiers advance automatically based on cumulative credit purchase thresholds.
Acceleration limits
Sharp usage spikes can trigger 429s independent of tier limits — ramp gradually.

Sources