Pricing

You pay for the task-seconds that you use. Computalot meters your tasks at its rate for the capacity class that you request. Computalot derives this rate from the live market rate for that capacity. Each task-attempt bills a minimum of 60 seconds. Every submit response includes summary.billing_estimate, the authoritative quote for that job, with the rate and the minimum itemized. Computalot releases the unused hold at settlement.

There are no subscriptions, minimum commitments, or idle charges. Credits are prepaid. A job starts only against a funded balance, so an open-ended bill cannot occur.

Indicative rates

Computalot prices capacity per machine-hour from a live market, so rates move with supply. The metered rates below come from live capacity that Computalot observed on 2026-07-12, and we refresh them periodically. They are orientation, not a rate card. The billing_estimate in your submit response governs your job.

Multi-GPU capacity scales almost linearly per GPU. You select capacity with requirements (cpu, memory_mb, gpu_count, gpu_memory_mb) or a named preset from GET /api/v1/presets. The platform fills the request from live market capacity and meters what your tasks use. The table shows the metered rate, which is the rate that you pay.

Worked examples

These examples use mid-range metered rates with the minimum applied. Your submit response shows the real numbers before anything runs.

1. Fan-out data processing (CPU). 1,000 tasks × 30 seconds each on 16-core capacity bills at the 60-second per-task minimum: ~16.7 billed machine-hours. At a metered rate of ~$0.20/hr that is ≈ $3.40 for the whole batch.

2. Hyperparameter sweep (GPU). A sweep job with 9 parameter combinations × 20 minutes each on 24 GB GPU capacity is 3 GPU-hours. At a metered rate of ~$0.47/hr that is ≈ $1.40.

3. Training run (datacenter GPU). 4 hours on one 80 GB GPU at a metered rate of ~$4.05/GPU-hr is ≈ $16.

How billing works

This is the one-minute version. The full contract is in Billing.

1. Fund — prepaid credits, topped up with USDC through x402 or MPP ($0.01 – $10,000 per top-up). Agents can do this autonomously.
2. Submit — the response carries summary.billing_estimate. A refundable hold is reserved, and it covers the initial attempt plus your max_retries budget. Underfunded submits get 402 with a ready-to-pay shortfall quote. Fund the account and retry the same request.
3. Run — usage is metered per task-second while tasks run. timeout_s caps the runtime cost of each task. Queue time is never charged.
4. Settle — at terminal state, the metered charge is captured and the unused hold is released automatically. The charge is the metered cost of your task-seconds, with the 60-second per-attempt minimum. billing_settlement.pricing itemizes this charge. In the rare case where the charge is more than the hold, the ledger records the difference as an overrun debit.

Other numbers that matter:

• Project init is free. It requires $5.00 of available balance as a funding floor.
• Artifact storage and transfer have no separate charge today. Retained artifacts count against a 100 GiB per-account quota until deletion.
• Infrastructure failures (lost workers, OOM kills) requeue without your retry budget. You do not pay extra for our misses.

Cost visibility

• GET /api/v1/account/balance — credits, holds, available funds.
• GET /api/v1/account/ledger — every settled credit and debit.
• GET /api/v1/jobs/:id — per-job billing_estimate, billing_hold, and post-terminal billing_settlement (actual cost, captured, released) inside summary.

Pricing FAQs

Can a job overspend my balance? No. Computalot admits a job only when the estimated hold fits your available balance. The hold bounds the reservation up front.

What happens to the unused part of the hold? It is released automatically at settlement. You keep it as available credits.

Is queue time billed? No. Only task runtime is metered.

What exactly makes up my rate? Computalot derives your rate from the live market rate of the capacity that your tasks run on. The rate also covers the orchestration layer: scheduling, retries, artifact plumbing, and the warm capacity that starts your tasks fast. Your settlement breakdown itemizes the charge.

Why a 60-second minimum per task? Very short tasks cost more to schedule, place, and account for than their raw seconds. The minimum keeps tiny fan-outs viable without a per-task fee. A 30-second task bills as one minute.

Do prices change? Yes. Capacity is market-priced, and the table on this page is therefore indicative. Your per-job billing_estimate is computed at submit time. That is the number to trust.