Skip to main content

Quota & rate limits

One bill, per organization

Platform usage is metered per organization, not per key — every key your org holds draws down one shared monthly budget and they rate-limit together. One key, one bill: AI, maps, and design all debit the same org meter.

The budget resets on a monthly calendar cycle.

What your plan includes

API access comes with a paid plan; your organization's plan sets the shared monthly token budget:

PlanMonthly API tokens
Pro5,000,000
Enterprise50,000,000

The Free plan doesn't include API keys — upgrade at kaleidr.com/billing. Plan changes apply to your existing keys within a few minutes; no re-mint needed.

When you hit the cap

A request that would exceed the cap returns 429 Too Many Requests with a body describing the meter and limit:

{ "detail": { "meter": "b2b.tokens", "tier": "…", "limit": 5000000 } }

For streaming endpoints the cap is also surfaced mid-stream as a quota SSE event so your UI can react before the stream ends.

Concurrency

Streaming endpoints are governed by a per-key concurrency slot — a burst of simultaneous streams beyond the slot budget is shed with a 429 rather than queued. Retry with backoff.

Handling it

  • Treat 429 as "slow down / upgrade," not "broken." Back off and retry.
  • Surface the mid-stream quota event in your UI (remaining budget).
  • See Errors for the full status table.