Desktop local AI softwareRun local models
Run local models
with LM Studio
Discover, download and run open models from a desktop interface, then expose them through local OpenAI- and Anthropic-compatible APIs.
macOS 14+Windows x64 / ARMLinux x64 / ARM64GGUFMLX on Apple Silicon
Is LM Studio right for you?
LocalClaw's practical summary, based on the current official documentation.
InterfaceVisual
Runs locallyYes
Recommended memory16 GB+
Will it run on my computer?
Model size and context still determine the real RAM and VRAM requirement.
Apple SiliconExcellent fitmacOS 14+, M1 through M4; MLX and llama.cpp runtimes
Windows x64 / ARMSupportedAVX2 required on x64; 16 GB RAM and 4 GB dedicated VRAM recommended
Linux x64 / ARM64SupportedAppImage; Ubuntu 20.04+ documented
Intel MacNot currently supportedUse another runtime or supported machine
What LM Studio provides
Model discovery and downloads
Search Hugging Face models and manage local files from the app.
Local model server
OpenAI-compatible Responses, Chat Completions, Completions and Embeddings endpoints.
Tool use and MCP
Tool calling support depends on the selected model and its chat template.
Using OpenClaw?
LM Studio can serve the local model. LocalClaw adds native installation, health checks and OpenClaw controls around the complete stack.