AI News

Claude API

Anthropic API 8 limits verified 2 Aug 2026

Quota reset forecaster

You hit a wall. When do you get it back?

time to reset
window opened quota returns

Published limits

PlanLimitWindow When exhaustedEvidence
Start 1,000 per minute Opus 5 — requests token bucket 429 with retry-after header ● Documented
Anthropic API rate-limit reference, fetched 2 Aug 2026.
Start 2,000,000 per minute Opus 5 — input tokens token bucket 429 with retry-after header ● Documented
Cached reads do not count toward ITPM on Opus 5, so effective throughput is far higher than the headline number.
Start 400,000 per minute Opus 5 — output tokens token bucket 429 with retry-after header ● Documented
Measured on tokens actually generated; max_tokens does not count against OTPM.
Build 5,000,000 per minute Opus 5 — input tokens token bucket 429 with retry-after header ● Documented
Anthropic API rate-limit reference, Build tier.
Scale 10,000,000 per minute Opus 5 — input tokens token bucket 429 with retry-after header ● Documented
Anthropic API rate-limit reference, Scale tier.
Start $500 per calendar month Monthly spend cap calendar month API usage pauses until next month ● Documented
Spend caps are separate from rate limits: Start $500, Build $1,000, Scale $200,000.
Build $1,000 per calendar month Monthly spend cap calendar month API usage pauses until next month ● Documented
Anthropic API rate-limit reference.
Scale $200,000 per calendar month Monthly spend cap calendar month API usage pauses until next month ● Documented
Custom tier has no monthly spend cap.