Free · v1.6.7

Download LumaBrowser

Local AI chat, image generation, agents, and a real browser they can drive. Everything is in the free download. No account, no API key, no usage limits, and nothing to upgrade to.

Windows .exe installer macOS (Apple Silicon) .dmg installer macOS (Intel) .dmg installer Linux .AppImage
or install from the command line
One command, any OS
$ npx lumabrowser start
Requires Node 18+ · Downloads the right build for your OS on first run and caches it under ~/.lumabrowser/

What you get, all of it, for nothing

There is no feature held back for a paid plan, because there is no paid plan. This is the whole product.

Is there a paid version?

Not right now.

LumaBrowser used to list Pro and Enterprise tiers. We removed them. The honest reason is that we would rather have people using this than have a pricing table that nobody was convinced by, and we do not want to charge for something until we are confident it is worth paying for.

If you are deploying LumaBrowser headless, in Docker, or across a fleet of machines and you need something we do not currently offer publicly, get in touch. We would genuinely like to hear what the requirement is.

Questions

Is LumaBrowser really free?

Yes. Every feature is in the download above: local AI chat, image generation, sub-agents, the knowledge base, Code Mode, browser automation, the REST API, and the MCP server. There is no paid tier, no trial timer, no account, and no usage limit.

Do I need an API key?

No. A guided setup downloads a local model sized to your hardware and everything runs on your own machine with no key at all. If you would rather use OpenAI or Anthropic models, paste your own key and pay them directly. We never mark up tokens.

What hardware do I need?

The setup wizard reads your hardware and picks a model that fits, so a modest laptop gets a small model and a workstation GPU gets a large one. NVIDIA cards use CUDA, other GPUs use Vulkan, and Apple Silicon is supported natively. The browser, automation, and API features work on any machine regardless of GPU.

What leaves my machine?

A lightweight version check-in (app version and platform, so we can count active installs) and generalized template selector maps that make the shared Template Builder repository smarter. User-specific context, input values, and session tokens are stripped. There is no third-party analytics or crash reporting in the app, and your API keys never leave your machine. Full details here.

What is the fastest way to install?

If you have Node 18 or newer, run npx lumabrowser start. The launcher fetches the right build for your OS on first run and launches the app; later runs start instantly. You can also npm install -g lumabrowser and then run lumabrowser start. Prefer desktop shortcuts and a normal install? Use the native installer above.

What changed in the latest release?

See what's new in 1.6 for release notes, behaviour changes, and early-access features.