Documentation Index
Fetch the complete documentation index at: https://developers.chameleon.io/llms.txt
Use this file to discover all available pages before exploring further.
To maximize your Experience, Chameleon sets very high limits on the number of total requests and the amount of concurrency we support. At this time enforce rate limiting as such:
- Global across all endpoints.
- User Profile Searching with
max_concurrent=1or in situations where the API is being used abusively. - Creating a Delivery with
max_concurrent=1per User Profile to handle the enforcement of the maximum Pending Deliveries.
- The
globalstrategy is applied to all requests with limited exceptions. Currently set to 60-120 requests per minute. - The
max_concurrentstrategy is used to limit based on the maximum concurrent requests. When limiting with this strategy, we subtract and add from an N-sized bucket. We give anX-RateLimit-Limitheader with N number of concurrent requests when that limit is reached. Please wait until at least one of the outstanding requests finishes. This strategy will queue for up to 4 seconds to allow other concurrent requests to complete. If no request completes during that initial 4 second wait, a 429 response will be sent. - The
bucketstrategy is used to limit based on the total number of requests. Using this strategy, the bucket window expires after N seconds. We give anX-Ratelimit-Waitheader with N number of seconds. This is the number of seconds to pause for until requests can proceed without immediate rate limiting.
Contact us to speak about Changing these limits.
| Strategy | Header | Example value | description |
|---|---|---|---|
| global | X-Retry-After | 19 | Up to N requests can occur per minute. Wait for at least 19 seconds before sending your next request. |
| max_concurrent | X-Ratelimit-Limit | 4 | Up to N requests can be “running” at the same time. Wait for at least 1 request to complete before retrying. |
| bucket | X-Ratelimit-Wait | 12 | Up to N requests can have started within the time window. Pause for N seconds before retrying your request. |

