LM Studio
Desktop application for discovering, downloading, and running large language models locally on personal computer hardware
It takes care to curate a safe software space. Help us do more.
&
| + | Local Model Inference free | Runs open-weight language models entirely on-device using llama.cpp and Apple’s MLX engine, allowing chat and API use without an internet connection or third-party servers. |
|---|---|---|
| + | OpenAI-Compatible Local Server free | Exposes a REST API compatible with OpenAI’s chat completions and responses formats on localhost, letting existing OpenAI SDK-based tools point at a local model with no code changes. |
| + | Hugging Face Model Browser free | Lets users search and download GGUF and MLX model files, in multiple quantization levels, directly from Hugging Face inside the app, so they can match file size to their hardware before downloading. |
| + | Headless Daemon (llmster) free | Installs a GUI-free daemon via a single shell command for servers, cloud instances, and CI pipelines, enabling automated local-inference deployments without a desktop environment. |
| + | MCP Server Support free | Connects local and remote Model Context Protocol servers to loaded models, allowing models to call external tools and data sources directly from a chat session. |
| + | Document Chat (RAG) free | Retrieves relevant passages from attached PDFs, text, or Word documents locally at inference time, so users can query long documents without sending their content to any server. |
| + | Speculative Decoding free | Supports draft-model and multi-token-prediction based speculative decoding, speeding up token generation for compatible models to shorten wait times during local chat. |
| + | Multi-Device Model Routing (LM Link) free | Lets a Bionic agent on one device dispatch inference to models loaded on up to five other LM Studio devices on the network, extending usable local compute across multiple machines. |
| + | Bionic Agentic Workspace free | Adds an agent that edits documents, writes and runs code, browses the web, and controls the computer using local or cloud open models, letting users delegate multi-step work instead of a single chat reply. |
...6 more features/limitations. Contact us to get a complete list of features and system requirements.
System Requirements
| # | Minimum | Recommended |
|---|---|---|
| 1 | Windows 10 (x64) or Windows 11 (x64/ARM, incl. Snapdragon X Elite) | |
| 2 | AVX2 instruction set required (x64) | |
| 3 | 16GB | |
| 4 | 4GB dedicated VRAM |
| # | Minimum | Recommended |
|---|---|---|
| 1 | macOS 14.0 (Sonoma) or newer | |
| 2 | Apple Silicon (M1/M2/M3/M4); Intel Macs not supported | |
| 3 | 8GB (small models and modest context only) | 16GB |
| # | Minimum |
|---|---|
| 1 | Ubuntu 20.04 or newer (x64 or ARM64/aarch64) |
| 2 | AVX2 support included by default on x64 |
Ratings
Not available, but we appreciate help! You can help us improve this page by contacting us.
Repository
LM Studio being a proprietary software, its source code is not publicly hosted.
License
Categories
Alternatives
Notes
- LM Studio (the desktop app) and LM Studio Bionic (its agentic mode for work/coding) are the same underlying product line from Element Labs, Inc.; Bionic is bundled into the free tier as of Aug 2026.
- Local use of the app has been free for both personal and commercial/at-work use since July 8, 2025, with no license form required; a metered cloud-credit tier and an Enterprise plan (SSO, model/MCP gating, internal endpoints) were added on top of the free local tier in 2026.