- Home
- AI Browsers
- MoltBrowser
MoltBrowser
Montréal, Canada
What it is
MoltBrowser is a free, open-source AI browser with an embedded on-device model and privacy features. Its page-aware AI can summarize articles, explain code, extract table data, and translate pages. The browser runs open models such as Llama, Mistral, Phi, and Qwen locally; its README also describes bringing personal keys for OpenAI, Anthropic, Gemini, and OpenAI-compatible endpoints. Its agent can click, scroll, navigate, and fill forms, and tasks can be scheduled to run automatically. The browser stores personal browsing memory on the device, while conversation history can be searched, exported or imported as JSON, or wiped in one click. MoltNet routes traffic over Tor and lets users select an exit country. MoltShield blocks ads, trackers, and YouTube ads, and dismisses cookie-consent banners. The maker says it scrambles device fingerprint signals and masks the user's IP address. Desktop builds are listed for macOS on Apple Silicon, Windows 10/11 x64, and Linux x64. iOS and Android are in development.
For and against
For
- Runs several open models locally
- Agent can navigate and fill forms
- Browsing memory is stored on-device
- Conversation history can be exported or wiped
- Free and open source under GPLv3
Against
- iOS and Android are still in development
- Listed builds are limited to desktop platforms
- Release versions differ by platform
The verdict
MoltBrowser pairs local AI and browsing automation with privacy controls and data portability. Its current listed builds are desktop-only, while mobile versions remain in development.
Who it suits
MoltBrowser may suit desktop users seeking local AI tools, browsing automation, and on-device memory. It is also relevant to users who want to inspect or contribute to an open-source project.
Plans and prices
Free forever
Free forever · fully open source
Read 7 Oct 2026 · geneye.ai
On the record
- Ad and tracker blocking
- MoltShield blocks ads, trackers, and YouTube ads using EasyList/uBlock filters and auto-dismisses cookie-consent banners.geneye.ai · read 7 Oct 2026
- AI models
- It runs open models such as Llama, Mistral, Phi, and Qwen locally, and its GitHub README also documents bring-your-own-key support for OpenAI, Anthropic, Gemini, and OpenAI-compatible endpoints.github.com · read 7 Oct 2026
- AI tools
- Its page-aware AI can summarize articles, explain code, extract table data, and translate pages.geneye.ai · read 7 Oct 2026
- Automation
- Its agent can click, scroll, navigate, and fill forms, with tasks that can be scheduled to run automatically.geneye.ai · read 7 Oct 2026
- Available desktop builds
- Download builds are listed for macOS on Apple Silicon, Windows 10/11 x64, and Linux x64; the README notes that current release versions differ by platform.github.com · read 7 Oct 2026
- Data portability
- AI conversation history can be searched and exported or imported as JSON, and stored data can be wiped in one click.geneye.ai · read 7 Oct 2026
- Extensions
- The product page advertises full Chrome extension compatibility; the GitHub README says it supports standard web-extension APIs and browser.ai.* APIs.geneye.ai · read 7 Oct 2026
- Fingerprint protection
- The maker says MoltBrowser scrambles device fingerprint signals and masks the user's IP address.geneye.ai · read 7 Oct 2026
- License
- The GitHub README identifies the source code license as GPLv3.github.com · read 7 Oct 2026
- Local memory
- MoltBrowser stores a personal memory of browsing activity on the device.geneye.ai · read 7 Oct 2026
- Mobile status
- The product page lists iOS and Android as in development.geneye.ai · read 7 Oct 2026
- Network privacy
- MoltNet routes traffic over Tor and lets users select an exit country from the menu bar.geneye.ai · read 7 Oct 2026
- Privacy
- The maker says local AI conversations stay on the computer and browsing memory is stored on-device.geneye.ai · read 7 Oct 2026
- Product
- MoltBrowser is an open-source AI browser with an embedded on-device model and built-in privacy features.geneye.ai · read 7 Oct 2026
- Support and contribution
- The project README welcomes contributions and says issues and pull requests are welcome.github.com · read 7 Oct 2026
