Rate limits

Each plan includes a monthly request allowance. Every API shares the same tiers — the plan you're on sets your volume, and one key works across every endpoint.

Monthly request caps

Plan	Requests / month	Best for
Free	100	Start building, no card required.
Starter	2,000	For small apps and side projects.
Pro	10,000	For production workloads.
Scale	50,000	For high-volume integrations.

Beta: only the Free plan is live today — paid tiers are marked “Coming soon” on the pricing page. The caps above are what each plan will grant.

How counting works

One successful HTTP request = one unit against your allowance, no matter which endpoint you call.
The counter is per calendar month (UTC) and resets at the start of each month.
Cache-friendly by design: repeat lookups within an endpoint's cache window are cheap for us and fast for you — but each HTTP call you make still counts.
Track your live usage anytime in your dashboard.

What a call costs

Your plan counts HTTP calls, and every call counts the same. What changes is the work behind it — which is why fresh=1 is rate-limited separately.

Request	Counts against your plan	What happens behind it
Within the cache window	1 call	Served from the stored snapshot. No upstream request, no classification.
Cache expired	1 call	The last good copy is served immediately and the refresh runs in the background, so you do not wait for it.
`?fresh=1`	1 call	Forces the upstream fetch and the classification before answering. Subject to a five-minute minimum between successful refreshes per subject and to a cooldown after a source failure — both reported through `Retry-After` and the `refreshDeferred` fields, so a deferred refresh answers with data rather than an error.

Calling fresh=1 in a loop does not get you fresher data: inside the five-minute window it returns the same reading and the response says so. Poll the cached endpoint instead.

When you hit the limit

Once you reach your monthly cap, further requests return 429 Too Many Requests with a message noting the limit. Requests succeed again after the monthly reset, or immediately if you upgrade to a higher tier. See Errors for the full response shape.

Need more volume? Compare plans and upgrade anytime.

Monthly request caps

Plan	Requests / month	Best for
Free	100	Start building, no card required.
Starter	2,000	For small apps and side projects.
Pro	10,000	For production workloads.
Scale	50,000	For high-volume integrations.

Beta: only the Free plan is live today — paid tiers are marked “Coming soon” on the pricing page. The caps above are what each plan will grant.

How counting works

One successful HTTP request = one unit against your allowance, no matter which endpoint you call.

The counter is per calendar month (UTC) and resets at the start of each month.

Cache-friendly by design: repeat lookups within an endpoint's cache window are cheap for us and fast for you — but each HTTP call you make still counts.

Track your live usage anytime in your dashboard.

What a call costs

Your plan counts HTTP calls, and every call counts the same. What changes is the work behind it — which is why fresh=1 is rate-limited separately.

Request	Counts against your plan	What happens behind it
Within the cache window	1 call	Served from the stored snapshot. No upstream request, no classification.
Cache expired	1 call	The last good copy is served immediately and the refresh runs in the background, so you do not wait for it.
`?fresh=1`	1 call	Forces the upstream fetch and the classification before answering. Subject to a five-minute minimum between successful refreshes per subject and to a cooldown after a source failure — both reported through `Retry-After` and the `refreshDeferred` fields, so a deferred refresh answers with data rather than an error.

Calling fresh=1 in a loop does not get you fresher data: inside the five-minute window it returns the same reading and the response says so. Poll the cached endpoint instead.