Real Chrome on Windows
A real graphics stack, real fonts, a real profile — not a headless container pretending to be a desktop. Sites see what they expect to see.
Pricing
A physical node is one real Chrome browser on our own Windows machines, reserved for the month and never metered by the hour. $10 each, and every allowance grows with the count.
Jump to
One physical node is one browser running at a time, on hardware we own. Every node carries the same allowances, whichever machine it lands on.
$10/ month
1 physical node × $10 a month
7 days money back · No metered hours · Cancel any time
Card payments are not yet available. Create an account and request the number of physical nodes you need. We confirm your order and the price above, then activate your nodes and send an invoice. Nothing is charged until you approve the order. Request a full refund within 7 days of your first payment.
What a physical node is
The same thing however many you buy. Nothing below is a bigger plan; it is what one physical node carries.
A real graphics stack, real fonts, a real profile — not a headless container pretending to be a desktop. Sites see what they expect to see.
Hold a session for six hours or six days, navigate it, refresh it, clear it, destroy it and start another. Browser hours are never metered, so the bill does not move when the work does.
A monthly egress allowance across datacenter, ISP, residential and mobile pools. Use your own proxies instead and that traffic does not count towards it.
Managed challenges are waited out. Simpler ones are solved by our own free solvers, which cost you nothing and are not counted. Anything those cannot do goes to a paid solving provider, and a monthly allowance of those comes with every node you buy, pooled across the account. Past it, the rate below applies.
One token header and a JSON body. Screenshots, rendered DOM, network capture and scripted multi-step runs, all on the same API.
POST /api/usage returns the limits on your key and what you have spent against them this period. Nothing is estimated and nothing waits for the invoice.
Where a physical node runs
One kind of node, at one price. Which of our machines runs it is our problem, which is the point of buying capacity rather than servers.
Machines in our own racks, on our own egress lines, all running the same Windows image and the same Chrome build — so a browser looks the same wherever it lands, and none of it is somebody else's elastic container billed by the second.
Each session is placed on a machine that is free when you start it, and it stays there for as long as that session lives. Nothing moves a running browser and nothing restarts it under you.
There is one thing to buy: a physical node, $10 a month, on one API key.
Every limit, per node
These are the exact values written onto your API key. The API enforces them, and POST /api/usage reports what you have used against them.
| Limit | One physical node | Your 1 node |
|---|---|---|
| Grows with every node the right-hand column is your whole account, on one API key | ||
| Concurrent browsers | 1 | 1 |
| Requests / second | 5 | 5 |
| Outbound data (egress) a month everything your browsers download, pooled across every proxy pool and every node on the account | 40 GB | 40 GB |
| Sessions / month | 5k | 5k |
| API calls / month | 100k | 100k |
| Paid challenge solving the free ladder is never counted | 300 / mo | 300 / mo |
| The same however many you buy per session, and per node, whatever the account holds | ||
| Browser hours | Unmetered | |
| Longest session | No limit | |
| Idle session reclaimed after | 30 min | |
| Steps per scripted run | 100 | |
| Proxy pools | All of them, chosen automatically | |
| Your own proxies | Included, and never counted | |
| Machine your session lands on | Whichever of our own machines is free | |
Beyond your allowance
Your browsers keep working when an allowance runs out; the extra usage is billed at the rates below. Nothing is charged by surprise: we email you before anything over an allowance is invoiced.
The lines are ours, so bandwidth is a cost we pass on rather than a second product — well under the $5–12 a gigabyte the market charges.
$0.003/ solve
Only solves that reach a paid provider are counted. The free ladder runs on every request at no charge, and a challenge we simply wait out costs nothing at all.
$10/ node / month
Add or remove nodes on your billing page and the ceilings move with them the same day — rate limit, egress, sessions and API calls. No upgrade, no migration, no new key.
Above the slider
Past the top of the slider, capacity stops being a number you click to and becomes a conversation: dedicated machines, private pools, your own egress lines and annual terms.
More physical nodes than the slider offers, your own egress, your own terms.
Before you buy
What the reservation covers, what happens when a session or an allowance ends, and how a change to your nodes is billed.
Create an account and ask for the nodes you want. We activate them, issue your API key, and invoice you at the price on this page — there is no card form to fill in and nothing is charged before you have agreed it. When card payment opens you move onto it from your billing page, at the same price.
Yes, from your billing page, whenever you like. The rate limit, the egress allowance and the monthly guards move with the count the same day, on the key you already have — there is no upgrade path to follow and no key to re-issue. Removing nodes works the same way.
No, and the reason is the same one the price is built on: a node is hardware we own and power whether it is busy or idle, so a free node is a node somebody else cannot have. What you get instead is 7 days of money back — the whole product, at full size, on your own workload. Ask within 7 days of your first payment and we refund it in full, no interview. The live browser on the home page is there to try before any of that, and it needs no account.
Because that is what the hardware actually is. Our fleet is machines we own, where an idle browser costs exactly what a busy one costs; renting elastic cloud compute by the second is somebody else's cost structure. Charging you for hours would mean charging for something we do not pay for, and it would make you architect around a meter instead of around your problem. Buy four nodes, keep them open all month, and the bill does not move.
Whichever of ours is free when you start a session. They are machines we own and run, in our own racks and on our own lines, all on the same Windows image and the same Chrome build. Every node carries the same allowances and costs the same, so there is nothing to choose and nothing to pay more for. If your work needs particular hardware pinned to you — a named set of machines, a private pool, a graphics card you can rely on — that is an Enterprise conversation and we are happy to have it.
The slider goes to 20. Past that it is worth a conversation rather than a slider: at that size we talk about dedicated machines, private pools, your own egress lines and annual terms — tell us what you are running and we will size it with you.
A node is a reservation, not a named machine: it is the right to have one browser running at a time, for the month. Each time you start a session we place it on a machine that is free, and it stays on that machine for as long as the session lives — a running browser is never moved, and nothing restarts it under you to make room for somebody else. What a machine failure does, you find out immediately: the next call on that session answers 404 rather than pretending.
As long as you keep using it. There is no maximum age on a paid node — hold a session for a week if the work needs it — but a session with no API calls for 30 minutes is reclaimed, because an idle browser is a node somebody else could be using. Any call resets that timer, and /api/sessions is enough on its own.
When a session does end — you destroyed it, the idle timer took it, or the machine failed — what was in the browser goes with it: cookies, tabs, the profile. Nothing else is lost, and the node is not: start another session and you are running again on the same reservation, at once. The pattern worth designing for is the one every long job wants anyway — keep what matters in your own store, and re-start and re-navigate when a call comes back 404. A scripted run does that work in one call rather than thirty.
A rate limit answers 429 with a Retry-After header. A monthly allowance answers 429 with the time the period resets. No free browser answers 503, and you are told whether that is the fleet being full or your own ceiling. Nothing is dropped silently and none of it costs you anything.
Every call on your token. A scripted run of thirty steps is one call, which is both the cheapest and the fastest way to do a lot in one session — the steps run on the node instead of making thirty round trips.
Yes, at no charge and with no bandwidth counted against you. Pass a proxy on start or switch it mid-session and it is used as-is. The managed pools are what a node includes.
POST /api/usage with your token returns the limits in force, this period's usage and when it resets. The same numbers are on your usage page in the members area, against the allowances your nodes carry.