LM Studio ​

LM Studio is a desktop application for discovering, downloading, and running LLMs on your own hardware. Its visual model browser lets you filter by size, quantization, and benchmarks before downloading. Once loaded, models run through a polished chat interface with configurable parameters. MCP plugs external tools into this local-first experience.

LM
LM Studio
lmstudio.ai · lmstudio.ai
TRANSPORT
Streamable HTTP ✓
PLATFORM
Windows · macOS · Linux
MCP VIA
Settings

Visual Model Discovery + MCP Tools ​

LM Studio's model browser is its standout feature. Search HuggingFace directly from the app, filter by parameter count and quantization level, check community benchmarks, and download with one click. Models load instantly into the chat interface.

Once running, MCP tools extend what the model can do. Local inference handles reasoning and generation, while MCP provides live data from your external systems — a clean separation of concerns.

Highlights:

  • Model browser — search HuggingFace, filter by size/quant, view benchmarks
  • One-click download — models download and load without manual setup
  • Chat interface — adjustable parameters: temperature, top-p, context length
  • Local server — OpenAI-compatible API for third-party app integration
  • Multi-model — load multiple models simultaneously, compare responses
  • GPU acceleration — automatic Metal (macOS), CUDA (NVIDIA), Vulkan detection
  • GGUF support — optimized for quantized model formats

How to Connect ​

1. Create a Token ​

In Vinkius Cloud, go to your server → Connection Tokens → Create. Copy the URL.

2. Add MCP Server ​

In LM Studio → Settings → MCP → Add Server. Paste your URL.

3. Chat with Tools ​

Select a model, start a conversation. MCP tools are available for external data queries.

FAQ ​

Can I compare how different models use MCP tools? Yes. LM Studio can load multiple models simultaneously. Ask the same question to different models and see how each handles MCP tool calls.

Which model formats does LM Studio support? Primarily GGUF (quantized) models from HuggingFace. Both CPU and GPU inference are supported.

Does LM Studio expose a local API with MCP? LM Studio's local server provides an OpenAI-compatible API. MCP tools configured in the app are accessible through this API.

Is LM Studio free? Free for personal use. Commercial licenses available for enterprise deployments.