Browser Use hosting

An agent that drives a real browser to finish real tasks.

Search & web agentsMITfrom $18.99/mo

In short

Browser Use hosting at AIHostBox means a dedicated, single-tenant Browser Use instance on our infrastructure: we install it, pin it to a reviewed release, put it behind TLS on your domain, back it up nightly and keep it patched. Plans start at $18.99 per month and the admin account, the data and any API keys are yours.

Real browserVision or DOMSession persistenceHeadless

Browser Use is an MIT-licensed Python library from the Browser Use team that gives a language model control of a real Chromium browser. The agent reads the page’s interactive elements, and optionally a screenshot, then decides where to click, what to type and when a task is done. You describe the goal in plain language, such as “download last month’s invoices from this portal”, and receive the result as text or structured output you define.

It differs from scraping libraries because it does not need selectors written in advance, and from the company’s own cloud service because the browser, cookies and data stay on a machine you control. Running Chromium alongside the agent is memory-hungry and fragile on a laptop that sleeps, changes networks and shares a browser profile with your personal accounts, so a dedicated server with its own profile is the cleaner home for it.

Browser Use hosting plans

ProRecommended minimum

8 GB RAM · 4 vCPU · NVMe

$18.99/mo

Order

MaxLarge document sets

12 GB RAM · 6 vCPU · NVMe

$27.99/mo

Order

Every plan is a dedicated single-tenant instance with NVMe storage, nightly off-box backups, TLS and managed patching. Quarterly, semi-annual and annual terms are discounted at checkout. Model tokens are not included — you bring your own key.

How we run Browser Use

We run the open-source browser-use Python agent together with a headless Chromium it controls over the Chrome DevTools Protocol, exposed as a service on port 8000, with a persistent user_data_dir so profiles survive restarts.

What we configure for you

Your provider key, such as OPENAI_API_KEY or ANTHROPIC_API_KEY, in the environment
headless Chromium with a persistent user_data_dir outside the application directory
storage_state files for sites you want the agent to reach while logged in
allowed_domains set per task where you tell us the sites involved
Chromium dependencies and fonts kept patched with the rest of the system

Which plan fits

Pro (8 GB, 4 vCPU) comfortably runs one or two agents, each with its own browser, which covers most scheduled or on-demand tasks. Max (12 GB, 6 vCPU) is for several agents in parallel or heavy pages such as dashboards and web apps that hold a lot of memory. The model itself runs at your provider, not on the instance.

The honest limits

Sites with aggressive bot detection or CAPTCHAs will often block a plain headless browser, and we do not add proxies or CAPTCHA solvers. Agents make mistakes on long or ambiguous tasks, so verify important actions such as purchases or submissions rather than trusting the agent’s own success report.

What people run Browser Use for

Legacy system automation

Drive an internal tool that will never get an API.

Structured collection

Gather data from sites that resist scraping.

Browser tools for your own agents

Run Browser Use as an MCP server so Claude or another MCP client can hand browser tasks to it.

Scheduled form work

Fill in the same supplier portal or government form on a timetable, with logins kept in a saved browser state.

Licence and version policy

Item	Position
Upstream licence	MIT
Sold as managed hosting	Yes
Tenancy	Single tenant — one customer per instance
Patching	Critical advisories within 24 hours, high severity within 72

Our full position on licences, including the applications we decline to sell, is on the licensing page.

Works well with Browser Use

More Search & web agents we host

Questions

Browser Use hosting: your questions

Is this allowed on any website?

That is between you and the site’s terms of service. Our acceptable use policy prohibits using it for credential stuffing, bypassing anti-bot protection or scraping personal data at scale.

Which models work with Browser Use?

The library has wrappers for OpenAI, Anthropic, Google Gemini, Azure OpenAI, AWS Bedrock, DeepSeek, Groq, Mistral, OpenRouter and Ollama, plus LiteLLM for anything else. Browsing is a demanding multi-step task, so capable models with vision give noticeably better results than small local ones. You bring your own key; tokens are billed by your provider.

How do I let it log in to sites without handing the password to the model?

Use saved browser state: log in once, export it with export_storage_state, and start later sessions from that file. For credentials the agent must type, the sensitive_data option passes placeholders to the model and substitutes the real values only in the browser. Combine it with allowed_domains so those values are never used on other sites.

Can I stop the agent visiting sites it should not?

Yes. The browser accepts allowed_domains and prohibited_domains lists with wildcard patterns, and navigation outside the allowlist is refused. We recommend setting an allowlist for every recurring task, both as a safety net against prompt injection on untrusted pages and to keep an agent from wandering.

Is this the same as the Browser Use Cloud API?

No. Browser Use Cloud is the company’s hosted service with its own API, stealth browsers, CAPTCHA solving and proxies, billed per use. We host the open-source library, which runs the agent and a normal headless Chromium on your instance. It does not include Cloud features such as CAPTCHA solving or residential proxies.

Run Browser Use without running a server

A dedicated Browser Use instance, deployed and maintained by us. Month to month, cancel whenever.

Browse the catalogue Talk to a human