A bill, not a meter.
Pick a plan by how many runs you want at once. Minutes are included, and your plan price is the whole bill.
Three terms, before the table
The hard cap is on by default
When your minutes run out, new runs are refused with quota_exceeded and a retry time. You don’t get a bill.
We never bill our own failures
A run that breaks because of the harness, the fleet or the network is marked error and costs nothing. It shows as credited.
Your price is held for 12 months
Published prices may go up. If you’re subscribed, yours doesn’t for a year, and you get 60 days’ notice of any change after that.
Plans
Free
$0
Try it on your own app.
- Concurrent runs
- 1
- Device minutes
- 200/mo
- Evidence kept
- 7 days
Indie
$12/mo
One engineer and their agent.
- Concurrent runs
- 2
- Device minutes
- 2,000/mo
- Evidence kept
- 30 days
Studio
$49/mo
A small team, every PR checked.
- Concurrent runs
- 4
- Device minutes
- 8,000/mo
- Evidence kept
- 90 days
Scale
$149/mo
The front of the queue.
- Concurrent runs
- 10
- Device minutes
- 30,000/mo
- Evidence kept
- 180 days
US dollars, billed monthly. Signing up needs an invite while this is a beta.
Have an invite? Sign up
Everything in each plan
| Feature | Free $0 | Indie $12 | Studio $49 | Scale $149 |
|---|---|---|---|---|
| Concurrent runs | 1 | 2 | 4 | 10 |
| Device minutes / month | 200 | 2,000 | 8,000 | 30,000 |
| Device minutes / 24 hours² | 60 | 400 | 1,500 | 5,000 |
| Evidence retention | 7 days | 30 days | 90 days | 180 days |
| Seats | 1 | 2 | 8 | 25 |
| API keys | 1 | 3 | 20 | Unlimited |
| Queue priority | Standard | Priority | High priority | Top priority |
| macOS build minutes¹ | — | 200 | 1,000 | 4,000 |
| PR comment integration¹ | — | Included | Included | Included |
¹ Not available yet. The hosted service runs builds you upload; building from source and PR comments come later.
² A rolling 24-hour limit, so a runaway agent loop can’t spend the month overnight. Not a pricing term.
The small print, in large print.
What is a device-minute?
A minute a device spends on your run: installing, launching, running the steps, taking screenshots, and the reset afterwards. Time in the queue isn’t counted. Every run reports the breakdown, so you can check the arithmetic.
Why is concurrency the thing you charge for?
Because it’s the scarce thing. Minutes cost us almost nothing; a booted device waiting for you is the actual constraint on a small fleet. So plans step on how many runs you can have at once.
How can $12 be real when BrowserStack starts at $249?
We own the Macs. A rented cloud Mac costs $150–1,100 a month; ours cost about $44 a month over three years, power included. We don’t sell 30,000 real handsets, either — just simulators and emulators, which is what an agent checking a change needs.
Is Free slower than paid?
Not once a run starts: it gets the same devices and runs to the end. The only difference is the queue. When several runs are waiting for a device, paid plans are picked first, so at a busy moment a free run may wait a little longer to start.
What happens when I hit the cap?
New runs are refused with a clear error saying which limit, how much you’ve used and when to retry. Runs already going finish. You’re never charged more than your plan.
How do I get into the beta?
Join the waitlist and tell us what you build and which agent you use. We send invites in small batches as the fleet grows, and you sign up with the code. The docs show exactly what you’ll be doing once you’re in.
Start on Free. No card.
Request beta access