Ottili AI API — Rate limits & quotas

Rate limits & quotas

The Ottili AI API applies two related protections: rate limits* (a short-window throttle that protects shared

infrastructure) and quotas* (the plan-scoped allowance your company can consume per period). Both are returned on

every response so your client can self-regulate instead of guessing.

Rate-limit headers

Every response — successful or throttled — includes standard rate-limit headers:

RateLimit-Limit: 600
RateLimit-Remaining: 597
RateLimit-Reset: 1752364800

The exact numeric ceiling depends on your plan and API key scopes. Always read the live values from these headers rather than hardcoding them — they may change as plans evolve.

Quotas and free vs paid usage

Your company's quota* is the total allowance your plan grants for a period (for example, requests per day and the

AI credit pool). How the two relate:

• Free / included allowance* — every plan ships an included monthly quota (request quota + included AI credits). It

resets at the start of each billing period and unused included amounts do not roll over.

• Paid plans* — raise both the request quota and the included credit pool, and unlock higher rate-limit ceilings.
• Credits* — AI requests are metered against the shared company credit wallet. See

[Credits & usage](/docs/ottili-ai-api-credits). When the credit balance or monthly budget is exhausted, generation

stops with a clear insufficient_balance (HTTP 402) or rate_limit_exceeded

(HTTP 429) error instead of over-spending.

The live numbers for your company are always shown in Settings → Billing* in the dashboard and in the response

headers above. Do not poll the generation endpoint to discover your limits — watch the headers and react to limit

errors.

When you are limited

When a rate limit is exceeded the API responds with HTTP 429 Too Many Requests, a Retry-After header telling you

how many seconds to wait, and the RateLimit-* headers reflecting the reset:

HTTP/1.1 429 Too Many Requests
Retry-After: 12
RateLimit-Limit: 600
RateLimit-Remaining: 0
RateLimit-Reset: 1752364860

{
  "detail": {
      "code": "rate_limit_exceeded",
    "message": "Too many AI API requests. Please retry after the indicated delay."
  },
  "code": "HTTP_429",
  "request_id": "b1c2d3e4-0000-1111-2222-333344445555"
}

The request_id is present on every response and is the handle to share with support (see

[Errors → Support](/docs/ottili-ai-api-errors)).

Backing off (retries)

Honour Retry-After exactly, then add a small jitter so many clients do not retry in lockstep. Exponential backoff

without Retry-After is a safe fallback:

import time, requests

def complete_with_retry(url, headers, body, max_retries=4):
    for attempt in range(max_retries):
        r = requests.post(url, headers=headers, json=body)
        if r.status_code != 429:
            return r
        wait = int(r.headers.get("Retry-After", 2 ** attempt))
        time.sleep(wait)
    raise RuntimeError("Rate limited after retries")

Authenticated, scoped requests receive higher ceilings than anonymous traffic. Cache static metadata and model lists

(GET /v1/models) to stay well under the limit, and send an Idempotency-Key on retried POST calls so a

retry never double-charges (see [Errors → Idempotency](/docs/ottili-ai-api-errors)).