PyPI · Rate Limits
Pypi Rate Limits
PyPI does not publish a hard per-second rate limit. The JSON, Index, and RSS APIs are heavily cached and CDN-fronted via Fastly, so PyPI states there is "currently no rate limiting of PyPI APIs at the edge." The legacy XML-RPC API may be rate-limited if usage degrades service. PyPI reserves the right to ban irresponsible consumers.
2 Limits
Throttle: 429
Package RegistryPythonOpen SourceRate Limiting
Limits
JSON / Index / RSS APIs IP
no edge rate limit (heavily cached via Fastly CDN)
PyPI states "currently no rate limiting of PyPI APIs at the edge"; consumers should still avoid thousands of requests per minute and should prefer ETag-conditional requests.
XML-RPC API (legacy) IP
rate-limited at PyPI's discretion if usage degrades service
PyPI may temporarily or permanently throttle XML-RPC consumers whose usage causes service degradation.
Policies
User-Agent Identification
Set a descriptive User-Agent that uniquely identifies your requests and ideally includes contact information; opaque or default agents may be rate-limited or blocked.
ETag and Conditional Requests
The JSON, RSS, and Index APIs return ETag headers — use If-None-Match conditional requests to avoid hammering uncached origins.
Caching Awareness
Inspect X-Cache and X-Cache-Hits to understand whether your requests are being served from CDN cache; design crawlers around the CDN rather than the origin.
Discretionary Bans
PyPI reserves the right to temporarily or permanently prohibit a consumer based on irresponsible activity; sustained high-volume crawling without identification or backoff risks a block.
Pacing Guidance
PyPI advises against thousands of requests in a short window (minutes); prefer serial requests over longer periods when feasible.