Rate limits

Quotas by tier

Both limits are enforced on a rolling window. Exceeding either returns 429 quota_exceeded with a Retry-After header (seconds until the next slot frees up).

Free-tier data limits

On top of the call quotas, the free tiers (Sandbox and Student) carry two data-volume limits that keep the API open for evaluation and learning without enabling bulk archive extraction. Paid tiers (Starter and above) are not subject to either.

• Requesting history older than your plan's window returns 422 window_out_of_plan with allowed.earliest — the window is not silently truncated.
• A single response above the per-call row ceiling returns 422 per_call_row_cap_exceeded. Narrow the window or add filters.
• The rolling 30-day row total is cumulative across calls. Exceeding it returns 429 export_cap_exceeded with used, cap, and resets_at; capacity frees as older usage ages out of the window.

Response headers

Every 2xx and 429 response includes:

X-RateLimit-Limit-Minute: 60
X-RateLimit-Remaining-Minute: 58
X-RateLimit-Limit-Day: 10000
X-RateLimit-Remaining-Day: 9842
X-Tier: starter
Retry-After: 12           # only on 429

• Limit fields are constant for a given tier — useful as a sanity check but you can hard-code them.
• Remaining fields are the authoritative source of truth. Always read them instead of counting requests client-side.
• X-Tier tells you which plan billed the request. Helpful in logs when you upgrade and want to confirm the key picked up the new tier.

Handling 429s

Back off based on Retry-After. Exponential backoff with jitter is the safest default:

import time, random, requests

def get(url, key, attempt=0):
    r = requests.get(url, headers={"X-API-Key": key}, timeout=10)
    if r.status_code == 429 and attempt < 5:
        delay = int(r.headers.get("Retry-After", 2 ** attempt))
        time.sleep(delay + random.random())
        return get(url, key, attempt + 1)
    r.raise_for_status()
    return r.json()

Burst behavior

The per-minute limit is a hard ceiling — bursts above it are rejected immediately, not queued. The per-day limit is informational until exhausted, then also hard-rejects.

If you need sustained throughput above your tier's per-minute limit, upgrade to the next tier or contact sales for a bespoke limit.