Browsergridby Zyxio
Documentation

Overview

Use the service

Billing

Reference

Documentation

Browsergrid docs

Captures, browser sessions and analytical jobs over one API, paid from prepaid credit. Start with the quickstart; every call of the grid has its own reference page, with what it costs and when it settles.

Overview

  • OverviewBrowsergrid captures web pages, lends you remote browsers and runs your containers, billed from credit or by monthly invoice per published result, browser-hour and vCPU-hour.
  • QuickstartFrom an API key to a published capture, a browser session, an analytical job and an MCP client.

Use the service

  • Managed servicesRequest PostgreSQL + PgBouncer, MLflow or LiteLLM; follow setup, obtain credentials and manage your instance.
  • CapturesSend up to 1,000 URLs per batch; each is fetched or rendered once and published as an immutable set of files you download with a short-lived URL.
  • Browser sessionsLease a remote Chromium, drive it over the Chrome DevTools Protocol, and pay per browser-hour only while it is ready.
  • Analytical jobsRun a linux/amd64 container image with the CPU and memory you choose, follow it through its phases and read its output.
  • MCPGive an AI agent captures, browser sessions and analytical jobs as tools, through the Model Context Protocol endpoint /mcp.
  • API keysCreate, cap, rotate and revoke the keys your programs use to call the grid.

Billing

  • Credit and billingYou pay in euros, from credit or by an invoice after each month against a spending limit you choose: work reserves what it may cost when it is admitted and is charged for what it actually used when it settles.
  • Pricing unitsWhat each unit measures, how a charge is computed and rounded, and worked examples at the prices of rate card 2026-10-10.sandbox-1.

Reference

POST /batchesSubmit up to 1,000 URLs as one capture batch, all under one script and one browser profile.
GET /batches/{batchId}A batch's progress counts and every item's state, with the item ids you need to fetch results.
GET /jobs/{jobId}One capture item or one analytical job: its state, attempts and failure reason.
POST /jobs/{jobId}/retryRun a failed capture item or a failed analytical job again.
GET /artifacts/{jobId}/{kind}A read-only download URL, valid for 15 minutes, for one published file of a capture item.
POST /sessionsLease a remote Chromium under one browser profile and get its one-time CDP address.
GET /sessionsYour live browser sessions, with their state and attestation, without their CDP addresses.
DELETE /sessions/{sessionId}Release a browser session: stop its meter and remove the browser.
POST /jobsSubmit an analytical job: a container image, a command and the resources it may use.
GET /jobsYour analytical jobs, newest first, filtered by phase or by one label, a page at a time.
POST /jobs/{jobId}/cancelAsk an analytical job to stop; what already ran is charged, the rest is released.
GET /jobs/{jobId}/logsThe stdout and stderr of one attempt of an analytical job, as plain text.
GET /jobs/{jobId}/artifactsThe files an analytical job has published: today, one log per finished attempt.
GET /gridThe grid's state right now: the capture queue, running workers and session slots.
GET /grid/queue-depthOne number: how many capture items the whole grid has queued or in progress.
POST /mcpThe Model Context Protocol endpoint: every grid call as a tool, over stateless Streamable HTTP.
GET /healthzWhether the grid's API is up; needs no credential and says nothing else.
GET /.well-known/oauth-protected-resourceThe grid's OAuth protected-resource metadata, which MCP clients read to find where to sign in.