Anthropic / Claude rate limit exceeded (429)
Number of requests has exceeded your rate limit
You’ve exceeded Claude’s requests-per-minute or tokens-per-minute limit for your tier. The API returns 429; back off, retry, and reduce burst load to recover.
What causes it
- Per-minute request or token limits for your usage tier.
- Large contexts pushing token-per-minute limits.
- Concurrent agents or retries spiking the call rate.
How to fix it
- Back off and retry with jitter; respect anthropic-ratelimit headers.
- Reduce concurrency and context size.
- Move up a usage tier for higher limits.
Your tier's limits
Anthropic rate limits by tier →
Requests and tokens per minute for each usage tier, and how to move up one.
Keep working on a fixed monthly plan
Standard Compute serves Claude-class and other frontier models behind one endpoint, retries transient upstream rate limits and can fall back to another provider for the same model, so a single provider’s limit stalls your workload less often.
Plans from $19 a month with $19 of compute. Eligible new accounts can try $0.25 of compute first, no card. Trial terms
FAQ
Does Claude have separate request and token limits?
Yes. Anthropic enforces requests per minute, input tokens per minute and output tokens per minute for each model on your tier, plus a monthly spend cap per tier. Large prompts hit the input-token limit first, though cache reads don’t count toward it on current models.