Overview
Local Studio is a local-first workstation for running, managing, and using self-hosted LLM backends on your Mac. It puts vLLM, SGLang, MLX, and llama.cpp behind a single OpenAI-compatible API, with a setup wizard that installs an engine, downloads a model from Hugging Face, launches it, and benchmarks it. A control surface shows live status, launch state, logs, GPU and runtime metrics, while the workbench combines models, providers, a browser, files, a terminal, and agent sessions in one place. The built-in agent runtime is powered by the Pi coding agent and keeps file operations local. Pair the KittyLitter companion app to continue the same sessions from an iPhone, iPad, or Android device while the work keeps running on your Mac. Free and open source under Apache-2.0.
Architecture: Apple Silicon
Key Features
- Runs vLLM, SGLang, MLX, and llama.cpp behind one OpenAI-compatible API
- MLX backend for native Apple Silicon serving
- Setup wizard installs engines, downloads models, launches, and benchmarks
- Live status dashboard with launch state, logs, and GPU/runtime metrics
- Workbench with models, providers, browser, files, terminal, and agents in one session
- Built-in agent runtime powered by the Pi coding agent
- Local-only agent file operations
- Hugging Face model downloads via built-in recipes
- Remote and LAN controllers with API key authentication
- KittyLitter mobile companion for continuing sessions on iPhone, iPad, or Android
- Free and open source (Apache-2.0)
