Home / Agents · Research / Fastbrowse
Fastbrowse
For web automation and data extraction: returns cited answers or schema-matched data.
What is Fastbrowse
Fastbrowse is an open-source browser agent that uses small models to choose clicks quickly and returns answers with citations or data matching a schema. Credentials and spending are controlled in code. It can be self-hosted or used as a hosted API with a free tier and paid plans from $29/month.
How to use
- Install uv, then set a model key such as OPENROUTER_API_KEY (or AI_GATEWAY_API_KEY).
- Run a task from the CLI with uvx fastbrowse "" --start ;
- add --local to use Chrome on your machine, --headed to show the window, or --profile to keep a signed-in profile.
- Pass credentials by name with --secret name=ENV_VAR so values never enter a prompt;
- add --authorize to allow irreversible actions such as placing an order.
- Read the result: a cited answer with quotes from the page, plus steps, model calls and cost;
- add --record to save a video.
- Or install the fastbrowse skill in Codex or Claude Code, or use fastbrowse-mcp from an MCP client, and hand your agent a website task.
- For the hosted version, create an account at console.fastbrowse.ai and send tasks to the API.
Use cases
- Giving a coding agent (Codex, Claude Code, Cursor) a browser for website tasks Automating web workflows such as logging in, filling forms and checking out, with answers backed by quoted page text Extracting schema-validated structured data from websites Using saved logins from Bitwarden or scoped secrets without exposing credentials to the model Running browser automation at scale on a hosted, sandboxed Chrome pool via API
FAQ
What is fastbrowse hosted?
The managed version of fastbrowse: you send a task to the API and get back a cited answer, or schema-validated data when you ask for it. The service runs the browsers, the model connections, the queue and the secret store. You can still run the open-source project yourself.
Do I need a card to start?
No. The Free plan needs no card and carries $2.00 of runs a month (at most 400 runs, since none costs less than $0.005), one run at a time, a 7 day history and a cap of 500 runs a month, top-ups included.
How is a run charged?
A run is a per-run fee plus usage, and no run is priced below $0.005. A plan's monthly allowance is spent first, then any top-ups. A platform error, a provider that was unavailable, a cancelled run, a run that expired in the queue and a run made with a test key are not charged.
What does a test key do?
A key beginning fbh_test_ runs a simulator with the same request and response shapes: no cost, and no website touched. That is the key to put in CI.
What happens when the balance runs out?
The API refuses the run with insufficient_balance and a 402, so nothing runs up a bill you did not expect. Top up, or turn on automatic top-up, before the next run.
Can I self-host instead?
Yes. fastbrowse is open source under the MIT licence, and the hosted service is a separate, optional product on top of it.
Price from official sources
Freemium
- fastbrowse is open source (MIT);
- fastbrowse hosted pricing: per-run fee plus usage, never below $0.005 per run.
- Free：$0, $2.00 of runs a month, 1 run at a time, 7-day history, 500 runs/month cap, no card.
- Starter：$29/month, $35.00 of runs included, 5 runs at a time, 30-day history, secrets, profiles, schedules, webhooks.
- Pro：$199/month, $250.00 of runs included, 25 runs at a time, 90-day history, session recordings, audit log and SSO-ready roles.
- Scale：$999/month, $1,400.00 of runs included, 100 runs at a time, 180-day history, queue priority, shared Slack channel.
- Enterprise：by agreement (dedicated browser fleet, custom retention and residency, invoiced billing, uptime commitment).
- Monthly allowances do not roll over;
- purchased top-ups do not expire.
Check the official site for current pricing · Price not yet checked by our editors
Source: fastbrowse.ai · Info as of Nov 2, 2026
Company
- Name
- Agent Labs, Inc.
Reviews
No reviews yet.
Alternatives
-
sweet-search
Free
For agents searching codebases: local semantic search that saves 5–33% per agent run
- Added
- Nov 2, 2026
Same category: Agents
-
Corollary
Free
For writing LaTeX on a Mac: an Overleaf-style editor powered by on-device models
- Added
- Nov 2, 2026
Same category: Agents
-
nonce
Not stated
For securely granting AI access to your email: issues time-limited, least-privilege access per question
- Added
- Nov 2, 2026
Same category: Agents
-
Skep
Free
For collaborating with coding agents across devices: share live sessions with a join code
- Added
- Nov 2, 2026
Same category: Agents
Social