LM Studio
#3 in Local LLM Runtimeslm-studio · since Mai 2023 (v0.1.x); aktuell v0.4.x (v0.4.0 am 28. Januar 2026) · 33× · last seen Jun 30, 2026
61
Momentum
LM Studio is a desktop application by Element Labs, Inc. for running open-source large language models locally on personal hardware without any cloud dependency. It provides a GUI, an integrated Hugging Face model browser, an OpenAI-compatible REST API, and a headless daemon (llmster) for server and CI deployments. Inference backends are llama.cpp (GGUF) and Apple MLX; the app supports macOS, Windows, and Linux, and has been free for both personal and commercial use since July 2025.
Momentum trend
04.04.03.07.
Features
| Deployment (Self-Hosted/Cloud) | Local (Desktop GUI or headless via llmster daemon); self-host on Linux servers, cloud instances, and CI; no cloud operation provided by LM Studio itself |
| Throughput/Latency | MLX on Apple Silicon ~30–50% faster than llama.cpp/Metal; Flash Attention for CUDA/Vulkan/Metal; continuous batching for parallel requests |
| License | Proprietary app license (Element Labs, Inc.); CLI (lms) under MIT license; free for personal & commercial use |
| Platform | macOS (Apple Silicon), Windows (x64/ARM64), Linux (x64, ARM); iOS/iPadOS via Locally app |
| Price | Free (personal & commercial); Enterprise plan for SSO, model gating, private collaboration available on request |
| Protocol Compatibility | OpenAI-compatible REST API (/v1/chat/completions, /v1/completions, /v1/embeddings, /v1/models, /v1/responses) on localhost:1234; Anthropic-compatible /v1/messages endpoint; MCP (Model Context Protocol) host; Python & TypeScript SDK |
| Release Date | May 2023 (v0.1.x); v0.4.0 on January 28, 2026 |
| Supported Models/Providers | GGUF (llama.cpp) & MLX models from Hugging Face; including Llama, Qwen, Mistral, DeepSeek, Gemma, Phi, gpt-oss; built-in model catalog (model.yaml) |