LM Studio – Discover and Run Local LLMs

LM Studio is a powerful desktop application for discovering, downloading, and running local large language models on your computer. With its intuitive graphical interface, LM Studio makes it easy to experiment with various open-source LLMs without any technical setup. It supports a vast library of models from Hugging Face and provides a local inference server compatible with OpenAI’s API.

Key Features

• Model Discovery – Browse and search thousands of models from Hugging Face
• One-Click Download – Download any compatible model with a single click
• Chat Interface – Beautiful built-in chat UI for testing models
• Local Server – OpenAI-compatible API server for application integration
• GPU Acceleration – Supports NVIDIA, AMD, and Apple Silicon GPUs
• Model Comparison – Run multiple models side-by-side
• Quantization Options – Choose between different quantization levels (Q4, Q5, Q8)

System Requirements

• RAM – 8GB minimum, 16GB+ recommended
• Storage – 10GB+ for application and models
• GPU – NVIDIA (CUDA), AMD (Vulkan), or CPU-only mode
• OS – Linux (Ubuntu 20.04+), Windows, macOS

Installation on Linux

# Download the AppImage from lmstudio.ai
wget https://lmstudio.ai/download/linux

# Make it executable
chmod +x LM-Studio-*.AppImage

# Run LM Studio
./LM-Studio-*.AppImage

Using the Local Server

LM Studio can run as a local API server:

# The server runs on localhost:1234 by default
# Compatible with OpenAI SDK

from openai import OpenAI

client = OpenAI(base_url="http://localhost:1234/v1", api_key="lm-studio")

response = client.chat.completions.create(
    model="local-model",
    messages=[{"role": "user", "content": "Hello!"}],
    temperature=0.7,
)

print(response.choices[0].message.content)

Popular Models to Try

• TheBloke Models – Optimized quantized versions of popular LLMs
• Llama 2 – Meta’s open-source foundation model
• Mistral/Mixtral – High-performance models from Mistral AI
• Phi-2 – Microsoft’s compact yet capable model
• Deepseek Coder – Specialized for code generation
• OpenHermes – Fine-tuned for helpful responses

Use Cases

• Model Exploration – Test different models before committing
• Private AI Chat – Secure conversations without cloud services
• Development – Local API for building AI applications
• Research – Compare model performance and capabilities
• Offline Work – AI assistance without internet

Download LM Studio