Pricing
Priced per run, not per seat.
Everyone on your team can see every run. You pay for the work, and the free tier is big enough to be honestly useful.
Solo
£0Free for good, up to 100,000 runs a month
For a side project, or the first month of a real one.
- 100,000 runs a month
- 3 schedules
- 7 days of history
- Community support
Team
£49per month, then £0.30 per 10,000 extra runs
For a product with customers who notice when work does not run.
- 2,000,000 runs a month
- Unlimited schedules
- 30 days of history
- Concurrency keys and priority lanes
- Email and Slack alerts
Scale
£249per month, volume pricing above 20M runs
For platforms running other people's work as well as their own.
- 20,000,000 runs a month
- 12 months of history
- Per-tenant isolation and quotas
- 99.95% uptime commitment
- A named engineer on a shared channel
Compare
Every limit, side by side
| Limit | Solo | Team | Scale |
|---|---|---|---|
| Runs included | 100,000 | 2,000,000 | 20,000,000 |
| Schedules | 3 | Unlimited | Unlimited |
| Run history | 7 days | 30 days | 12 months |
| Max job duration | 5 minutes | 30 minutes | 2 hours |
| Concurrency keys | — | Included | Included |
| Priority lanes | — | 3 | 10 |
| Dead letter queue | Included | Included | Included |
| Per-tenant quotas | — | — | Included |
| Uptime commitment | — | 99.9% | 99.95% |
Questions
The ones people actually ask
One attempt at one job. If a job fails and retries three times before it succeeds, that is four runs. Schedules count a run each time they fire, whether or not your endpoint answers. Runs you cancel before they are picked up are free.
On Team and Scale the extra runs are billed at the overage rate and nothing stops. On Solo, enqueueing is paused until the month rolls over, and your schedules keep their next fire time so nothing is lost, just late.
No. Vault Inc calls an HTTPS endpoint you own with the job name and arguments, and records what you return. Your code, your database and your secrets stay where they are, which is also why there is no runtime or language list to check.
Your job definitions are a YAML file in your own repository and your workers are ordinary HTTP handlers, so the thing you would have to replace is the queue and the scheduler, not your application. Run history exports as newline-delimited JSON.
Two months off if you pay for a year up front, on Team and Scale. It is applied at checkout rather than as a code, and switching back to monthly is prorated.
Start on the free tier and stay there as long as it fits.
No card, no trial clock, no sales call in between.