Tiers, quotas and limits

How compute credits work

Every engine meters the compute it actually consumes in credits: one credit ≈ one second of medium-weight compute (backend/core/credits.py). Your plan's monthly credit allowance is what you can spend that month; a weekly cap and a per-hour spend cap sit on top so a single runaway loop cannot drain the month at once. One-time credit packs raise the current month's allowance by their size. DEMO-labelled surrogate results never bill; failed real engine runs bill only the seconds they consumed before failing.

The four plans

• Free — full platform access, taste-sized compute allowance.
• Starter — for individual researchers running real experiments regularly.
• Pro — heavier campaigns, bigger datasets and priority queueing.
• Enterprise — self-serve shared managed capacity for lab-wide use (highest queue priority; OIDC/Google sign-in).

Every plan ships every module — synthesis, evaluation, prediction, optimization, the physics engines, the API and SDK. Tiers differ only in compute allowance, quotas, queue priority, support and the free-tier dossier watermark. The Pricing page renders the live numbers; API responses and Settings → Billing always govern what is enforced.

What is limited

• Compute allowance — monthly credits, plus the weekly cap.
• Jobs per hour — how many runs you can start in a rolling hour.
• Concurrent jobs — how many runs may execute at once.
• Data points per dataset — rows × columns budget.
• Synthetic rows per run — per-generation cap.
• Max upload size — per-file limit.
• Total storage — across all datasets and results.

What happens at 402 and 429

HTTP 402 means your credit balance is exhausted: top up a credit pack or upgrade a plan, then retry. HTTP 429 means a quota window was hit — jobs per hour, the weekly credit cap, the monthly allowance, daily wall-clock or concurrency. The response names the limit and when it resets. Neither is a permanent failure; nothing is silently dropped.

Where to see your numbers

Your tier and current usage appear in Settings → Billing and on the Runs page. The Pricing page documents every plan's allowance and limits publicly.

Fair use

These limits keep the platform responsive for everyone. Repeatedly exceeding them may result in a temporary rate limit or, for deliberate abuse, account suspension — see the Terms of use.