# Syzygy

> Turn your Mac into a private AI coworker. Run open LLMs locally — offline, no API keys — working across your apps, files, browser, and code.

## Products

- [Mach Studio](https://withsyzygy.com/product): The macOS desktop app: local models, connected apps, browser control, automations.
- [Mach](https://withsyzygy.com/mach): The Apple Silicon Mixture-of-Experts inference engine behind Mach Studio.
- [Mach Agents](https://withsyzygy.com/sdk): @syzygy-ai/agents — the TypeScript agent SDK built for small and local models.
- [Syzygy Gateway](https://withsyzygy.com/gateway): OpenAI- and Responses-compatible inference gateway; one key for every provider.
- [About](https://withsyzygy.com/about): Company mission and thesis.
- [Pricing](https://withsyzygy.com/product#pricing): Plan comparison.

## Documentation

- [CLI](https://withsyzygy.com/docs/mach/cli): mach-generate, mach-check, mach-models, and mach-archs.
- [Installation](https://withsyzygy.com/docs/mach/installation): Install mach from PyPI, add optional speculative decoding, and verify the setup.
- [Introduction](https://withsyzygy.com/docs/mach): Inference engine for Apple Silicon. Runs Syzygy Research Mach-1 packs with fused decode-in-GEMV Metal kernels and an OpenAI-compatible API.
- [Mach-1 Small benchmarks](https://withsyzygy.com/docs/mach/benchmarks): Methodology, result tables, and limitations for published Mach-1 Small quality and local-performance claims.
- [Serving](https://withsyzygy.com/docs/mach/serving): Run mach-serve and connect OpenAI or Anthropic clients, with streaming, tool calling, image input, and performance options.