Thunderbird/Thunderbolt
Summary
Thunderbolt is an open-source, cross-platform AI client that enables users to deploy on-premises with support for frontier, local, and on-prem models while maintaining data ownership and avoiding vendor lock-in. Currently in early active development with enterprise targeting, it supports web, iOS, Android, Mac, Linux, and Windows platforms.
View Cached Full Text
Cached at: 04/20/26, 08:33 AM
thunderbird/thunderbolt
Source: https://github.com/thunderbird/thunderbolt
Thunderbolt CI (https://github.com/thunderbird/thunderbolt/actions/workflows/ci.yml)
AI You Control: Choose your models. Own your data. Eliminate vendor lock-in.
Thunderbolt Main Dashboard
⚠️ We are excited about the amount of interest Thunderbolt has been getting and want to clarify that it is still early and under active development. Currently, we are targeting enterprise customers that want to deploy it on-prem. We encourage you to self-host it and try it out, but there are a few caveats we are still working on:
- While we eventually plan to make Thunderbolt fully offline-first, it currently depends on authentication and search functionality (though you can disable search on the integrations screen in the app). You can deploy your own backend with Docker and sign up in order to test it locally.
- You’ll need to add your own model providers - we don’t yet have a public inference endpoint. We recommend using Thunderbolt with Ollama (https://ollama.com) or llama.cpp (https://github.com/ggml-org/llama.cpp) if you want free local inference, or you can add API keys for any OpenAI-compatible model provider in the settings.
Thunderbolt is an open-source, cross-platform AI client that can be deployed on-prem anywhere.
- 🌐 Available on all major desktop and mobile platforms: web, iOS, Android, Mac, Linux, and Windows.
- 🧠 Compatible with frontier, local, and on-prem models.
- 🙋 Enterprise features, support, and FDEs available.
Thunderbolt is under active development, currently undergoing a security audit, and preparing for enterprise production readiness.
Need Help?
Found a bug? Have an idea?
- We’re actively working on our docs, community, and roadmap. For now, the best way to get in touch is to file an issue.
Contributing
We welcome contributions from everyone.
- Development: The development guide will help you get started.
- Make sure to check out the Mozilla Community Participation Guidelines.
Documentation
- FAQ - Frequently asked questions
- Deployment - Self-host with Docker Compose or Kubernetes
- Development - Quick start, setup, and testing
- Architecture - System architecture and diagrams
- Features and Roadmap - Platform and feature status
- Claude Code Skills - Slash commands, automation, and subtree syncing
- Storybook - Build, test, and document components
- Vite Bundle Analyzer - Analyze frontend bundle size
- Tauri Signing Keys - Generate and manage signing keys for releases
- Release Process - Instructions for creating and publishing new releases
- Telemetry - Information about data collection and privacy policy
Code of Conduct
Please read our Code of Conduct. All participants in the Thunderbolt community agree to follow these guidelines and Mozilla’s Community Participation Guidelines.
Security
If you discover a security vulnerability, please report it responsibly via our vulnerability reporting form. Please do not file public GitHub issues for security vulnerabilities.
License
Thunderbolt is licensed under the Mozilla Public License 2.0.
Similar Articles
Mozilla Thunderbolt AI: Run Your Own AI Agent and Keep Your Data Private
Mozilla's for-profit subsidiary MZLA Technologies launched Thunderbolt AI, an open-source, self-hosted AI client that runs entirely on user infrastructure, supports any model, and ensures data privacy by keeping data inside the organization's own systems.
Looks like PyTorch is getting fast thunderbolt communication backend (distributed models on Macs)
PyTorch is reportedly adding a fast Thunderbolt communication backend for distributed model training on Macs.
OWC Stack AI brings Thunderbolt 5 local AI support to Windows and Linux
OWC announces Stack AI, a Thunderbolt 5 device that combines AI acceleration with storage expansion to enable running larger local AI models on Windows and Linux, reducing cloud dependency and improving privacy.
@zcbenz: MLX's implementation of RDMA (Remote Direct Memory Access) over Thunderbolt on macOS, can now be used as an independent…
MLX's RDMA-over-Thunderbolt implementation for macOS is now available as a standalone library, enabling high-speed Mac clusters for local AI workloads.
@akshay_pachaar: Apple finally did it. Its new framework, Core AI, runs models entirely on Apple silicon, so inference happens on the us…
Apple released Core AI, a new framework that runs AI models entirely on Apple silicon devices (iPhone, iPad, Mac, Vision Pro) with zero server calls. It includes a memory-safe Swift API, model export recipes for PyTorch, an optimizer, and debugging tools, supporting models like Qwen, Mistral, and SAM3.