Documentation

Rate limits & quotas

Subscription keys run against a 24h sliding token window. Pay-per-token keys are unbounded except by your prepaid balance.

#Subscription quota

Each subscription tier has a 24h rolling token cap. Every subscription response carries the full quota + routing headers:

http

X-Powapi-Quota-Remaining: 1234567
X-Powapi-Quota-Reset-At:  1717000000   (unix seconds)
X-Powapi-Tier:            pro
X-Powapi-Model-Used:      deepseek-v3
X-Request-Id:             req_01J9...

When you hit the 24h limit you get a 429 with the same quota headers, plus a Retry-After hint.

#Concurrent streams

Each tier caps the number of in-flight streaming requests per tenant (Starter: 2, Pro: 5, Max: 12). Excess concurrent calls 429 immediately.

#Pay-per-token

No request-rate ceiling — usage stops only when your balance hits zero. Keep an eye on /tenant/credits or set up a low-balance alert in your dashboard.