Rate limits
How much you can send, and how to handle 429s.
These limits keep one runaway client from affecting everyone else.
| Limit | Value |
|---|---|
| Requests per key | 300 / minute |
| Requests in flight per account | 16 |
Beyond these, the limits are Anthropic's own, such as maximum tokens and context window for each model. Hitting a limit returns 429 with a retry-after header.
Handling 429s
retry.py
from anthropic import Anthropic
# The SDK already retries 429s and 5xx with backoff; raise the ceiling for heavy workloads.
client = Anthropic(max_retries=5)Need higher limits for a production workload? Email support@gethephaestus.com.