What this product does
Use LM Studio to find, download, and chat with a local model in a desktop app.
- Provides a desktop window for working with models
- Lets other apps use a model running on your computer
- Can run in the background without an open desktop window
- Helps you find, download, and open models
Requirements before downloading
The original desktop app requires Apple Silicon and macOS 14 or newer on Mac; Intel Macs are unsupported. Windows supports x64 with AVX2 and supported Snapdragon ARM systems. Linux provides x64/ARM64 AppImages; the vendor lists Ubuntu 20.04 or newer and AVX2 for its default x64 build. Check the current system requirements for your exact platform before choosing the installer.
The vendor recommends at least 16GB RAM; this is app guidance, not a guarantee that a particular model and context fit. Bionic and headless llmster are separate applications with their own setup requirements.
Where it fits in a local AI setup
- 1Your prompt
- 2LM Studio
- 3Local model
- 4Answer
Connecting other apps through an API
- Chat completions
- POST /v1/chat/completions: chat messages and streamed answers.
- Responses
- POST /v1/responses: supported Responses requests; compatibility is a subset of the hosted API.
- Legacy completions
- POST /v1/completions: text completion.
- Embeddings
- POST /v1/embeddings: requires an embedding model, not an arbitrary chat model.
- Model listing
- GET /v1/models: use a returned model identifier in requests.
- Decision models
- POST /v1/decisions: decision-model requests in supporting builds.
- Tool calling
- Tool-capable models can emit tool calls; the client executes them and returns results.
- Structured output
- JSON/schema requests still need a model and runtime supporting the selected schema. Validate the returned object.
- Streaming
- Request stream: true on a supporting generation endpoint.
Local base address: http://localhost:1234/v1. Use the port shown by your server. Endpoint availability does not establish every model supports every task. Enable the local server before connecting; authentication and available endpoints depend on its configuration.
First successful setup
On the official download page, choose the Download LM Studio section and the installer for your computer, then open LM Studio. The separately listed Bionic and llmster products use different interfaces.
In Discover search Qwen/Qwen3-0.6B-GGUF and download Qwen3-0.6B-Q8_0.gguf. This small model checks installation rather than demanding answer quality.
In My Models open the gear menu, set Context Length to 4096 and save the loading defaults. In Chat select and load that download.
Ask for a plain-English rewrite of “The meeting will commence at 10:30” that preserves the time. Check the reply, stop generation if necessary, then unload. Reopen the app and select the same download to use it again. If loading fails, close memory-heavy apps and reduce the context before retrying.
What success looks like: the model loads without memory pressure and returns a normal chat response.
Choose something else when
- a multi-user browser workspace: compare Open WebUI;
- an unsupported computer: check the requirements before downloading;
- agentic project work: compare the separate Bionic application and its execution choices.
License
The desktop terms permit personal and internal business use subject to their restrictions. Redistribution and hosted-service use are not implied. Downloaded models and hosted services have separate terms.
License: Proprietary desktop terms. Read the controlling software terms
Run without an open window
LM Studio can run through llmster, its standalone background service. If you already use the desktop app, its server-on-login setting keeps the service running after the window closes to the tray. Choose the desktop app for browsing models, or llmster for a machine that needs only the model service.