A browser your agent can use, and you can account for.

Hosted, isolated Chromium sessions that report what they did, what they consumed, and what you were charged, line by line.


What one session actually costs

Staged example
ses_8f2a41c9e07b Closed normally 2026-08-27 14:02:11 → 14:02:23 UTC
Elapsed session time Elapsed time in this example, distinct from billable usable-browser time. 12.4s
Billable browser time Billed per second, starting when the browser is usable. Example excludes no telemetry gaps. 12.4s
Provider compute-seconds Duration multiplied by CPU count. Shown for diagnosis. Not billable browser time. 24.8cpu·s
Measured traffic Egress recorded by the node. Decimal megabytes. 3.1MB
Stored files One screenshot, counted against your storage cap, not against this session's spend. 148KB
Idle time before close Not separately measured. Idle time inside an open session is billed as session time. not measured
Settled charge 0.07credits

You keep the stop. A session you opened is a session you can end, from the portal or the API, and billable time ends according to the recorded session lifecycle. Settled charges come from an immutable ledger: we never reprice history.

What we measure

  • Usable browser time, billed per second
  • Egress traffic, per session
  • Stored file bytes against your cap
  • Credits reserved for running sessions, with usage deducted as it accrues

What we don't claim

  • That a held credit is already spent, or a released hold is income
  • That compute-seconds are browser time
  • That an unknown measurement is zero
  • That returning from checkout means you were charged

Rendered, interactive, inspectable

A fetch tool returns the shell. A real browser runs the page's JavaScript, operates its controls, holds an authenticated profile between runs, reaches an application on your own machine through a tunnel, and leaves you a screenshot to check the work against. Each of those is a capability with limits, and the documentation states them.


Start with one useful result

Connect your agent, open a page, and keep a screenshot of the result.

If you are an agent: start with the API documentation; the shared skill download is coming soon.

browserious login
browserious --json session create
browserious page create SESSION_ID --url https://example.com
browserious screenshot page SESSION_ID PAGE_ID --output-file first.png
browserious session delete SESSION_ID

Use the session and page IDs returned by the preceding commands.