LM Studio — локальный запуск языковых моделей

★ —

LM Studio is a desktop app (Windows, macOS, Linux) for running large language models locally, offline, with no data sent to third-party servers. Reach for it when you want to work with open models on your own computer conveniently: find and download a model from Hugging Face, chat in the built-in interface, pick a quantization to fit your hardware, and, when needed, spin up a local server with an OpenAI-compatible API to plug the model into your own apps. Everything is private — data never leaves the machine. Its focus is a ready "download and run a model" app with a GUI, not a developer library or a cloud service: it runs on llama.cpp/MLX under the hood, so you need enough RAM and ideally a GPU. For a console/server scenario without a GUI people usually pick Ollama, and for a team web interface, Open WebUI.