Pricing units
What each unit measures, how a charge is computed and rounded, and worked examples at the prices of rate card 2026-10-14.commercial-1.
Units
| Price key | Unit | Price | In micro-euro |
|---|---|---|---|
capture.fetch | one published result | €0.12 per 1,000 | 120 per result |
capture.browser-fetch | one published result | €0.80 per 1,000 | 800 per result |
capture.render-fetch | one published result | €0.80 per 1,000 | 800 per result |
capture.capture | one published result | €1.00 per 1,000 | 1,000 per result |
session.browser | browser-hour, accrued per millisecond | €0.07 per hour | 70,000 per hour |
audit.base | unique URL checked | €0.22 per 1,000 URLs | 220 per URL |
audit.render | URL or state rendered, on top of its base check | €0.37 per 1,000 URLs | 370 per URL |
job.vcpu | vCPU-hour, accrued per millisecond | €0.04 per vCPU-hour | 40,000 per hour |
job.memory | GiB-hour, accrued per millisecond | €0.0064 per GiB-hour | 6,400 per hour |
runner.2vcpu | runner-hour, 2 vCPU, accrued per millisecond | €0.0036 per runner-minute | 216,000 per runner-hour |
runner.4vcpu | runner-hour, 4 vCPU, accrued per millisecond | €0.0071 per runner-minute | 426,000 per runner-hour |
postgres.plan | Tier-1 base workspace-month | €49.00 per workspace-month | 49,000,000 µ€ per month |
mlflow.plan | Tier-1 base workspace-month | €39.00 per workspace-month | 39,000,000 µ€ per month |
litellm.plan | Tier-1 base workspace-month | €29.00 per workspace-month | 29,000,000 µ€ per month |
redis.plan | Tier-1 base workspace-month | €15.00 per workspace-month | 15,000,000 µ€ per month |
The four managed service plans are live and self-service. See Managed services for what each Tier-1 base plan includes. The audit.* and runner.* prices are for offerings that are coming; see Coming, below.
Managed services come in five size tiers, tiny, starter, standard, pro and scale, with 1 to 8 replicas. standard is the Tier-1 base plan priced above. The other tiers' prices are proposed, tiny from €2.00 to €3.50 per month depending on the service, and no rate card carries them yet: until one does, an instance of any tier accrues its service's plan price above.
Micro-euro
All amounts are whole micro-euro (µ€): €1 is 1,000,000 µ€. The API's credit errors and the panel's statement use them so that no amount is ever rounded twice.
How a charge is computed
- Per result. A published capture item is charged its script's price, once.
- Per time. A session or a job attempt is charged its hourly rate times its duration in milliseconds, rounded half up to the micro-euro once, at settlement:
(rate per hour × milliseconds + 1,800,000) ÷ 3,600,000, keeping the whole part. - A job attempt's duration runs from its container's start to its finish and stops at
runningDeadlineSeconds. It counts only when the attempt ended for a reason you caused (your code's exit, your own memory limit, your own deadline, your cancellation); an attempt that ended for a reason on our side counts zero milliseconds. See Analytical jobs. - Never above the reservation for captures, sessions and jobs. Their charge never exceeds the part of the reservation set aside for that item, session or attempt.
- A job's rate is the CPU you request in vCPU times €0.04 plus the memory you request in GiB times €0.0064, per hour.
500mis 0.5 vCPU;512Miis 0.5 GiB.
Managed service proration
Plan charges accrue per active millisecond, using the actual length of each UTC calendar month. For a period within one month: monthly price × active milliseconds ÷ milliseconds in that month. A period crossing a month boundary is split by month; its fractions are summed exactly, then rounded half up once to the micro-euro for that settlement period. February, a 30-day month and a 31-day month therefore have different per-millisecond rates.
Only the access binding's active time counts. Suspended and retired time costs nothing; requesting suspension does not stop accrual until it completes. Provisioning does not reserve a full month up front. Each settled period records a closed reservation equal to its charge, with no ongoing hold. See Credit and billing for how plans affect credit and monthly limits.
Worked examples
| Work | Computation | Charge |
|---|---|---|
| 10,000 rendered fetches, all published | 10,000 × 800 µ€ | €8.00 |
| 10,000 rendered fetches, 300 failed | 9,700 × 800 µ€ | €7.76 |
| One browser, ready for 30 minutes | 70,000 µ€ × 0.5 hour | €0.035 |
| One browser leased for 30 minutes, released after 12 minutes of use | 70,000 µ€ × 0.2 hour | €0.014; the other €0.021 of the reservation is released |
| A 2 vCPU / 4 GiB job running 10 minutes | (2 × 40,000 + 4 × 6,400) µ€ × ⅙ hour | €0.0176 |
The same job with a 10-minute deadline, stopped at it (RUNNING_DEADLINE_EXCEEDED) | (2 × 40,000 + 4 × 6,400) µ€ × ⅙ hour | €0.0176, however late the stop |
The same job evicted after 10 minutes (PLATFORM_EVICTED) | 0 ms | €0 |
The same job's reservation with a one-hour deadline, whatever its retries | 105,600 µ€ × 1 hour, one attempt | €0.1056 reserved |
| PostgreSQL active for 15 days of a 30-day month | 49,000,000 µ€ × 15 ÷ 30 | €24.50 |
| Valkey / Redis active for one day of October (31 days) | 15,000,000 µ€ ÷ 31, rounded once | €0.483871 |
| A route audit (coming) that checks and renders 500 URLs | 500 × 220 µ€ + 500 × 370 µ€ | €0.295 |
Coming
Shown for planning; not offered yet: route audits (audit.base, audit.render) and Actions runners (runner.2vcpu, runner.4vcpu) at the prices above; the route audit's jev semantic add-on (audit.semantic) and the 8 vCPU Actions runner (runner.8vcpu), not priced yet; and web app hosting, which no rate card prices yet.
See Credit and billing.