# Atomic Chat

> Free local AI chat app for macOS, Windows, Linux, iOS and Android. Run 1,000+ open-source AI models privately on your device with TurboQuant: 2x larger context window and 30% less RAM usage. No cloud, no fees, no tracking. Open-source, Apache-2.0.

## Key Pages

- [Homepage](https://atomic.chat/): Download and get started.
- [Models catalog](https://atomic.chat/models): 1,000+ open models to run locally, with filters for reasoning, code, vision and tools.
- [llms-full.txt](https://atomic.chat/llms-full.txt): Complete details, technical specs, and comparisons.
- [Privacy Policy](https://atomic.chat/privacy): How we handle your data.
- [Terms of Service](https://atomic.chat/terms): Terms of use.

## Features

- [Offline AI](https://atomic.chat/offline-ai): Run AI without internet, everything stays on your device.
- [Private AI](https://atomic.chat/private-ai): No cloud, no account, nothing logged.
- [Local LLM](https://atomic.chat/local-llm): Run the best open LLMs on your own machine.
- [TurboQuant](https://atomic.chat/turboquant): KV cache compression, bigger models on 6x less RAM.
- [MCP connectors](https://atomic.chat/mcp-connectors): Wire local models to Notion, GitHub, Drive and 1,000+ tools.
- [Self-hosted LLM](https://atomic.chat/self-hosted-llm): Local OpenAI-compatible API without the server setup.
- [Unfiltered AI](https://atomic.chat/unfiltered-ai): Uncensored open models with no cloud filter.

## Platforms

- [Local LLM on Mac](https://atomic.chat/local-llm-mac): Apple Silicon at full speed.
- [Local LLM on Windows](https://atomic.chat/local-llm-windows): One installer, NVIDIA GPUs and plain CPUs.
- [Local LLM on Linux](https://atomic.chat/local-llm-linux): GUI or headless via the local API.
- [Local LLM on iPhone](https://atomic.chat/local-llm-iphone): Models run fully offline on Apple Silicon.
- [Local LLM on Android](https://atomic.chat/local-llm-android): AI in your pocket, no internet needed.

## Models

- [Qwen](https://atomic.chat/qwen): Qwen3, Qwen3.6 and Qwen2.5 builds.
- [DeepSeek](https://atomic.chat/deepseek): V4, V3, R1 and Coder builds.
- [Gemma](https://atomic.chat/gemma): Google's Gemma 4 and Gemma 3 builds.
- [Llama](https://atomic.chat/llama): Meta's Llama 3.1 and Llama 3.2 builds.
- [GPT-OSS](https://atomic.chat/gpt-oss): Open-weight 20B and 120B builds.
- [Kimi](https://atomic.chat/kimi): Moonshot's K2 builds for agentic coding.

## Blog

- [Blog](https://atomic.chat/blog): Guides and updates on running LLMs locally.
- [Guides](https://atomic.chat/blog/guides): Step-by-step local LLM setup guides.
- [LLM updates](https://atomic.chat/blog/llm-updates): Model releases and engine benchmarks.