Guides

Limits and concurrency

Requests per minute, requests at once, and how deposits raise your level.

Two limits apply to inference:

Limit	Scope	When you pass it
Requests per minute	Organization, and the key if it has its own	`429 rate_limit_exceeded`
Requests at once (concurrency)	Organization	`429 concurrency_limit_exceeded`

Concurrency grows with the total you have deposited over time. The levels are on the pricing page and your current level is in the console under Limits. Every response carries x-notholy-concurrency-limit and x-notholy-concurrency-remaining.

curl

curl -s -D - -o /dev/null https://api.notholy.com/v1/chat/completions \
  -H "Authorization: Bearer $NOTHOLY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "notholy-specter", "max_tokens": 16, "messages": [{"role": "user", "content": "ok"}]}' | grep -i x-notholy

Requests also reserve credits while they run. If the reservation does not fit in your available balance, the request gets 402 before it starts.

Research routes (/v1/cve, /v1/web, /v1/files) have their own per-minute limit.

If you need more than the top level, write to sales from notholy.com/enterprise.

Errors

Limits and concurrency

Next