Premier desktop suite for discovering, running, and serving local LLMs offline with Metal/CUDA hardware acceleration and built-in API server.
Completely free for personal learning and non-commercial research; includes unrestricted Hugging Face discovery, model downloading, and local API serving.
LM Studio is a premier desktop suite designed to run local LLMs offline on consumer laptops and workstations. Leveraging optimized backends for Apple Silicon (Metal) and NVIDIA GPUs (CUDA), it allows seamless discovery and downloading of GGUF quantized models directly from Hugging Face. LM Studio includes both an interactive chat playground and a built-in local OpenAI-compatible API server that integrates effortlessly with developer tools.