Rate limits

Call, token, size and wait limits, with the error each returns.

Every limit returns an error code. Retry-After is in seconds.

Limits

Limit	Value	Error	Reset
Calls per account key (`abk_`)	60 per minute and 5,000 per 24 hours by default; set per key	`429 rate_limited`	Rolling 1 minute and 24 hours
Chat completions per LLM key (`hlk_`)	200 per UTC day by default	`429 daily_limit`	00:00 UTC
Tokens per member account	300,000 per UTC day	`429 daily_tokens`	00:00 UTC
Open chat completions per member account	64	`429 too_many_streams`	When a completion ends
Open chat completions per account and model	The model's `max_parallel`	`400 too_many_parallel`; `400 parallel_unsupported` when `max_parallel` is 1	When a completion ends
Chat request body	32 MB	`413 too_large`	None
Tokens per chat request	The model's `context_length`	`400`	None
Media processing per member account	600 s per UTC day by default	`429 budget-exhausted`	00:00 UTC
Open media jobs per account	1,000	`429 too-many-open`	When a job ends
JSON request body on audio and media routes	4 MB	`413 too-large`	None
File in `POST /v1/media/uploads` or `POST /v1/audio/transcriptions`	20 MB	`413 too-large`	None

A member account is an albastov.com account other than the owner's. The owner's account has no token, open-completion or media-processing limit.
Account limits apply to every key of the account. max_parallel does not apply to hlk_ keys.
max_parallel and context_length per model are in GET /v1/models.
Open completions count toward the hlk_ daily limit; GET /v1/models does not.
A 429 daily_limit or 429 daily_tokens body carries limit and used; 429 budget-exhausted carries resets_at.

Limit	Value
Publish requests, per publish key	120 a minute, bursts of 30
Published events, per publish key	300 a minute; each event of a batch counts
Requests, per events key	600 a minute, 200,000 a day
Publish body	16 KB; batch 256 KB and at most 100 events
`idempotency_key` window	7 days, per source
Webhook body	2 MiB
Webhook deliveries, per source	120 a minute, bursts of 30
Webhook noise budget, per source	5 events a minute one by one; the rest as one summary event
Kept webhook deliveries, per source	The newest 200, bodies cut at 256 KiB
Open streams, per read key	8
Stream catch-up	1000 events
History	90 days per source by default

Condition	Response
Model not loaded	`503 model_loading` with `Retry-After` and `retry_after_s`; a streamed request stays open until the model answers
Request waiting in the queue	`503 queued` with `position` and `Retry-After`
Speech or transcription not finished in time	`503 queued` with `job_id` and `Retry-After`; the job stays in media jobs
Non-streamed request with no response within about 100 s	`524`, plain text; stream the request instead

Retry rules per status: Errors.