Harness Hub
All startups

Browser Use

The company behind the open-source web-agent library.

Browser Use maintains the eponymous 100k-star open-source library for vision-capable browser agents and commercializes it for enterprises automating web workflows.

Pros & cons

Each point cites its sources and says what kind of evidence backs it. A second reviewer tried to refute every claim before it was kept.

Pros

  • MetadataVery popular MIT library for LLM-driven browser agents, with about 119k stars.[1]
  • DocsWorks with OpenAI, Anthropic, Google and local Ollama models, as well as its own BU2 model.[1]
  • DocsCloud stealth browsers start at $0.02 per browser-hour, with proxies and captcha solving.[1]
  • DocsIn November 2025 the vendor reported cutting cost from 39.2 cents to 1.9 cents per task and average task time from 123s to 33s over a year.[1]

Cons

  • DocsThe team itself calls reliability "the biggest pain point preventing widespread adoption."[1]
  • User reportsA production user reports about 25-30 steps per site, each with one LLM call, taking 3-5 minutes per site, and was looking for alternatives.[1]
  • DocsBenchmark claims such as 90.5% on Stealth Bench V2 are self-published.[1]
  • Independent testToken usage grows with page context, which can dominate cost at scale.[1]
Best for
Developers prototyping browser automation agents in Python with any LLM.
Watch out for
Add deterministic fallbacks and budget for retries on brittle sites.

Assessed Oct 29, 2026 · adversarially reviewedSuggest a correction

People

Their open-source work

Featured

browser-use/browser-use

Open-source Python browser agent: the model writes Python in a persistent REPL that drives Chrome over CDP.

Maintained by Browser Use

1.0.7released
3 pros · 4 cons

Activity

Last push
· yesterday
Latest release
1.0.7
· yesterday
Last 90 days
20+ releases
  1. 1.0.8rc11.0.8rc1: release candidate for 1.0.8Pre-release

    · yesterday

    pip install 'browser-use==1.0.8rc1' The judge asks as one more turn of the run, so the provider reads the run from its cache instead of billing the trace again. It returns the same verdict fields, gotchas included (#6457). No API key over SSH or on a headless machine: the run…

  2. 1.0.7browser-use 1.0.7

    · yesterday

    pip install 'browser-use==1.0.7' MCP calls no longer time out at 60 s. A long browser_exec now replies within 50 s with the output so far, so Cursor and other clients with a 60 s limit get a result instead of Request timed out. The cell keeps running, and calling browser_exec…

  3. 1.0.6browser-use 1.0.6

    · 5d ago

    pip install 'browser-use==1.0.6' BU 4.0 is the default model. Agent(task) and ChatBrowserUse() now run bu-4.0 on the Browser Use gateway instead of bu-fast. Pass llm='bu-4.0-mini' for simple tasks at the lowest cost, or llm='bu-4.0-pro' for the hardest tasks (#6430). This costs…

  4. 1.0.5browser-use 1.0.5

    · 7d ago

    pip install 'browser-use==1.0.5' Every run is judged by your own model: one more low-effort request checks the run against the task and fills result.judgement with success, failure_reason, description, prompt_injection_detected, bot_block_or_captcha_detected and gotchas.…

  5. 1.0.4browser-use 1.0.4

    · 8d ago

    pip install 'browser-use==1.0.4' The agent drops old screenshots in one batch once 20 tool results carry one, instead of on every new screenshot. Each prune rewrote an earlier message and broke the provider's prompt cache, so screenshot-heavy runs paid full input price on…

Checked against GitHub on Nov 21, 2026. Release notes are shortened to plain text.

On X

No posts from them in our X feed yet.

Follow @browser_use on X · The X feed

In the news

This update brings remote access, on-device voice dictation, and browser use controls to GitHub Copilot for JetBrains. Copilot users in JetBrains IDEs also get context usage shortcuts, historical diffs, simpler… The post GitHub Copilot in JetBrains, Ignite releases appeared fi...

The integration layer: hundreds of managed tools your agents can call.

Composio handles auth, schemas, observability and reliability for the external SaaS tools an agent uses — 'plumbing for agentic actions' across Slack, GitHub, Salesforce and 250+ other surfaces.

4 pros · 4 cons

A registry and toolchain for MCP servers.

Smithery indexes and helps deploy MCP (Model Context Protocol) servers — a distribution layer for the tool ecosystem agents connect to.

3 pros · 4 cons

Exa

Search API purpose-built for AI agents.

Exa (formerly Metaphor) provides neural search and web-crawl APIs tuned for agent consumption — live, cited knowledge for agents instead of stale training data.

4 pros · 4 cons

Connections

Relations from the Harness Hub knowledge graph, as charted in the State of Harness reports.

Maintains · 2

Similar entries · 2

The Harness Hub newsletter

New harness companies, projects, releases and the monthly State of Harness report, in your inbox.