Free models
Models whose id ends in:free share one limit per account, at any balance. Every key and every free model on the account counts toward it:
Over the limit you get
429 rate_limit_exceeded with a Retry-After header in seconds. The message names the limit you hit and the wait, for example:
:free. Paid models have no request limit.
Paid models
Balance
While a paid request runs, NagaAI reserves 0.05 credits of your balance and releases the reserve when it charges the real cost. The number of paid requests that can run at the same time is your balance divided by 0.05, rounded down. With $2.00 you can run 40 at once. Below $0.05 a paid request fails with402 insufficient_quota.
Key spending limit
A key can carry its owncredit_limit, set on the API Keys page or through the API. When the key’s spend reaches the limit, its requests to paid models fail with 403 credit_limit_exceeded. Requests to free models still go through, under the free-model limit. NagaAI checks the limit before a request starts, so the last request can take the spend slightly past it.
Request size and time
NagaAI refuses a request body that is too large or arrives too slowly:Traffic from one IP address
More than 300 requests in 10 seconds from one IP address toapi.naga.ac are blocked for 10 seconds at the network edge, before they reach NagaAI.
Overload
Each NagaAI server runs a fixed number of requests at once. While it has room, any account can use as much of it as it needs. When the server is full, one account can hold at most half of its slots. That way one account’s burst can’t lock out everyone else. Your API keys and Playground sessions count together toward your account’s half. Once your account holds half of a full server, its next request fails at once. The answer is429 rate_limit_exceeded with Retry-After: 1. Requests that are already running are not affected. The message is:
/v1/messages the same refusal comes in the Anthropic envelope, with rate_limit_error as the type. To avoid it, cap how many requests your client runs at the same time and retry after Retry-After.
When the whole server is full, other accounts’ requests return 503 with Retry-After: 1. NagaAI sends no x-ratelimit-* headers, so read Retry-After to decide when to try again.