DevelopersEssentials
Rate limits and quotas
Each key has a per-minute rate limit and a daily quota, set by your plan. Every response tells you where you stand, and a request over the limit gets a 429 with how long to wait.
On this page
Limits by plan
| Key belongs to | Per minute | Per day (UTC) |
|---|---|---|
| Starter account | 60 | 10,000 |
| Professional account | 120 | 50,000 |
| Enterprise account | 300 | 200,000 |
| Partner (reseller) | 600 | 500,000 |
| Sandbox keys, per IP address | 30 | 2,000 |
- Limits are per key: two keys on one account each get the full amount.
- A partner key's limit covers everything it does, including work on clients through
AssistCall-Account. - The minute window is fixed: it resets at the start of each clock minute. The daily quota resets at midnight UTC. Requests refused with a 429 don't use quota.
- Your dashboard's Developers page shows your limits and a chart of requests and errors for the last 30 days.
Token endpoint limits
POST /v1/oauth/token has its own limits, so a broken loop can't lock you out of the rest of the API:
- 10 new tokens a minute per key (
token_rate_limited). A token lasts an hour, so you rarely need more than one. - 30 requests a minute to the token and revoke endpoints from one IP address (
rate_limited). - 10 failed attempts (wrong secret, bad or expired token) from one address within 10 minutes block that address for 10 minutes (
too_many_failed_attempts).
Rate limit headers
Every authenticated response carries the current minute window:
- RateLimit-Limit
- Requests allowed in the current window.
- RateLimit-Remaining
- Requests left in it.
- RateLimit-Reset
- Seconds until the window resets.
- RateLimit-Policy
- The policy, e.g.
60;w=60: 60 requests per 60-second window. - Retry-After
- On a 429 only: seconds to wait before trying again.
HTTP/1.1 429 Too Many Requests
Content-Type: application/json
Request-Id: req_4f1c9b2a7d3e8f60a1b2c3d4
RateLimit-Policy: 60;w=60
RateLimit-Limit: 60
RateLimit-Remaining: 0
RateLimit-Reset: 23
Retry-After: 23
{
"error": {
"type": "rate_limit",
"code": "rate_limited",
"message": "This key can make 60 requests a minute. Wait for the window to reset.",
"request_id": "req_4f1c9b2a7d3e8f60a1b2c3d4"
}
}When the daily quota is used up, the code is daily_quota_exceeded, and RateLimit-Limit, RateLimit-Remaining (0) and RateLimit-Reset describe the day instead, with Retry-After counting down to midnight UTC.
Retrying with backoff
On a 429, wait Retry-After seconds, then retry. For 5xx answers and network errors, wait a little longer each time (1, 2, 4, 8 seconds…) with some randomness, so many clients don't retry in lockstep. Give up after a few attempts and log the Request-Id.
# curl can retry 429s and 5xx itself, honoring Retry-After
curl --retry 5 --retry-max-time 120 \
https://api.assistcall.ai/v1/calls \
-H "Authorization: Bearer $ASSISTCALL_TOKEN"POST.Staying under the limits
- Use webhooks instead of polling. Let webhooks tell you when a call ends, then fetch just that call.
- Page with
limit=100and filters likestarted_afterrather than many small pages. - Reuse access tokens for their whole hour.
- Spread batch jobs out. If
RateLimit-Remaininggets low, pause untilRateLimit-Reset.
Need more? Email [email protected] with what you're building.