Parallel workers begin failing with rate_limit_error.
Why this happens
The application is exceeding an account or model limit. Immediate retries amplify the traffic spike.
Answer 1: Honor retry-after when it is present
import time
from anthropic import Anthropic, RateLimitError
client = Anthropic()
def create_with_retry(max_retries=5):
for attempt in range(max_retries):
try:
return client.messages.create(
model="your-enabled-model-id",
max_tokens=1024,
messages=[{"role": "user", "content": "Hi"}],
)
except RateLimitError as exc:
if attempt == max_retries - 1:
raise
headers = getattr(getattr(exc, "response", None),
"headers", {})
delay = float(headers.get("retry-after", 2 **
attempt))
time.sleep(delay)
