Guide · Open source

How to use web-ui: install, configure an LLM and run browser agents

Updated 2026-10-29 · written with AI from the project's own documentation and checked against our catalogue data

web-ui is an open-source, browser-based control panel for the browser-use project. It lets an AI agent operate a web browser to complete tasks on websites, and you set it up through a web page instead of writing code for each run. This guide walks through installing it locally or with Docker, configuring an LLM provider, and running your first agent task.

Browser Use Web UIbrowser-use/web-ui · ★ 16.6k · MIT Install options & downloads

Quick start

1Clone the repository. Run git clone on the web-ui repository and cd into the web-ui folder.
2Create the Python environment. Run uv venv --python 3.11 and activate the environment for your operating system.
3Install dependencies. Run uv pip install -r requirements.txt, then playwright install --with-deps to install the browsers.
4Configure .env. Copy .env.example to .env and add your LLM API keys and other settings.
5Start the WebUI. Run python webui.py --ip 127.0.0.1 --port 7788 and open http://127.0.0.1:7788.
6Run a task. Choose your LLM and browser settings in the page, enter a task for the agent, and watch it work.

What web-ui does and who it suits

You configure an LLM provider and browser settings in a web page, give the agent a task, and watch it work. The README says it is built on Gradio and supports most browser-use functionality. It lists Google, OpenAI, Azure OpenAI, Anthropic, DeepSeek and Ollama among the supported LLMs.

Good fit: people who want to try browser-automation agents without writing code, and developers comparing LLM providers.
Good fit: users who want the agent to use their own logged-in Chrome profile, or who want to watch sessions through a VNC viewer in Docker.
Not a fit: production systems that need hardened, audited automation. The README describes it as a UI layer and gives no production guidance.
Not a fit: anyone without an API key or a local model setup, since the agent needs an LLM to run.

Prerequisites before you install web-ui

For a local install you need Python 3.11, with uv recommended for environment setup, and Playwright browsers, which the setup commands install. You also need an API key for an LLM provider, or a local Ollama setup. Docker and Docker Compose are only needed for the Docker option.

Python 3.11 (uv recommended)
Playwright browsers, installed through the setup commands
An API key for an LLM provider, or a local Ollama setup
Docker and Docker Compose, only for the Docker option

Install web-ui locally with uv and Playwright

Clone the repository, move into the folder, and create a Python 3.11 environment. Then activate it. The activation command differs by operating system, so use the one that matches yours.

Windows (Command Prompt): .venv\Scripts\activate
Windows (PowerShell): .\.venv\Scripts\Activate.ps1
macOS/Linux: source .venv/bin/activate

git clone https://github.com/browser-use/web-ui.git
cd web-ui
uv venv --python 3.11
uv pip install -r requirements.txt
playwright install --with-deps

Configure your LLM keys in the .env file

Copy the example environment file to .env, then open .env in a text editor and add your API keys and other settings. This file is where the agent gets its LLM access, so setup involves editing it rather than clicking through a form.

Windows (Command Prompt) and macOS/Linux/Windows (PowerShell) both use the same copy command pattern shown in the README.
Leave BROWSER_USER_DATA empty if you want to use local user data.

cp .env.example .env

Run the WebUI and open it in your browser

Start the app with the command below, then open the address in your browser. The README uses 127.0.0.1 and port 7788, which you can change with the flags shown.

Open http://127.0.0.1:7788 in your browser.
Optional: use your own Chrome profile by setting BROWSER_PATH and BROWSER_USER_DATA in .env, closing all Chrome windows, opening the WebUI in a non-Chrome browser such as Firefox or Edge, and checking "Use Own Browser" in Browser Settings.

python webui.py --ip 127.0.0.1 --port 7788

Run web-ui with Docker Compose and VNC

The Docker option avoids a local Python setup. After copying .env and adding your settings, build and start the containers. On ARM64 systems such as Apple Silicon Macs, the README gives a variant command with TARGETPLATFORM set.

The Web UI runs on port 7788. The VNC viewer, for watching browser interactions, runs at port 6080. The README gives a default VNC password of "youvncpassword", which you can change by setting VNC_PASSWORD in .env.

Web-UI: http://localhost:7788
VNC viewer: http://localhost:6080/vnc.html

docker compose up --build
TARGETPLATFORM=linux/arm64 docker compose up --build

What people use it for

Trying browser automation AI agents without writing code.
Comparing LLM providers such as OpenAI, Anthropic, or Ollama for the same browser task.
Running an agent with your own logged-in Chrome profile so you do not need to re-login to sites.
Watching agent sessions live through a VNC viewer in a Docker setup.
Keeping the browser window open between AI tasks to review the history and state of past interactions.

Frequently asked questions

Is web-ui free to use?

web-ui is released under the MIT licence, as shown in its repository. You still need an API key for an LLM provider, or a local Ollama setup, because the agent requires an LLM to run.

What do I need to install web-ui?

You need Python 3.11, with uv recommended for the environment, plus Playwright browsers installed through the setup commands. You also need an LLM API key or local Ollama, and Docker with Docker Compose only if you choose the Docker option.

Can web-ui use Ollama instead of a paid API?

The README lists Ollama among the supported LLMs and describes a local Ollama setup as an option. Configure it through your .env settings and the WebUI settings page.

How do I use my own Chrome profile with web-ui?

Set BROWSER_PATH and BROWSER_USER_DATA in .env, close all Chrome windows, open the WebUI in Firefox or Edge, and check "Use Own Browser" in Browser Settings. The README notes this avoids re-logging into sites.

Why can't I see the browser when I run web-ui in Docker?

The Docker setup includes a VNC viewer at http://localhost:6080/vnc.html for watching browser interactions. The default password is "youvncpassword", and you can change it by setting VNC_PASSWORD in your .env file.

How does web-ui compare with browser-use, Skyvern, Stagehand or Browserbase?

web-ui is a Gradio-based UI layer built on the browser-use project. Skyvern, Stagehand and Browserbase are listed as comparable projects; the source does not compare their features, so check each project's documentation before choosing.

New agent skills, repos and AI tools — every week

The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.

Our verdict

web-ui suits people who want to try browser-automation agents through a web page, and developers who want to test different LLM providers or their own Chrome profile. Skip it for production automation, since the README gives no production guidance, and skip it if you have no LLM access or dislike editing a .env file.

Install options & downloads