Run AI models locally using LM Studio — a user-friendly desktop app with a graphical interface.
What is LM Studio?
LM Studio is a free desktop application for running AI models locally. It features a visual model browser, easy downloads, and a built-in local server — no command line needed.
Benefits:
- Easy GUI — no terminal required
- Built-in model browser — search and download models visually
- OpenAI-compatible API — works seamlessly with AIDA
- Free — no API costs, data stays on your machine
Step 1 — Install LM Studio
Download LM Studio from lmstudio.ai and install it on your computer.
Supported on: Windows, Mac, Linux
Step 2 — Download a Model
- Open LM Studio
- Click the Discover tab (magnifying glass icon)
- Search for a model — we recommend qwen2-0.5b to start
- Click Download
Recommended models:
| Model | Size | RAM Required | Speed |
|---|---|---|---|
| qwen2-0.5b | 531MB | 2GB | Very Fast |
| llama-3.2-1b | 1.3GB | 4GB | Fast |
| llama-3.2-3b | 2GB | 6GB | Medium |
| mistral-7b | 4.1GB | 8GB | Slow on CPU |
Step 3 — Start the Local Server
- Click the Developer tab (code icon on left sidebar)
- Click Local Server
- Select your downloaded model from the dropdown
- Click Start Server
You should see Status: Running and the server address http://localhost:1234.
Step 4 — Download the AIDA Launcher
The launcher creates a secure tunnel so AIDA can reach your local LM Studio server from anywhere.
Download the launcher for your OS:
- Windows: Download start-aida-lmstudio.bat (inside the AIDA app)
- Mac / Linux: Download start-aida-lmstudio.sh (inside the AIDA app)
You also need cloudflared in the same folder as the launcher:
- Windows: Download cloudflared.exe — rename to
cloudflared - Mac: Run
brew install cloudflared - Linux: Run
wget https://github.com/cloudflare/cloudflared/releases/latest/download/cloudflared-linux-amd64 && chmod +x cloudflared-linux-amd64 && mv cloudflared-linux-amd64 cloudflared
Step 5 — Run the Launcher
Make sure LM Studio server is running first, then:
- Windows: Double-click
start-aida-lmstudio.bat - Mac/Linux: Double-click
start-aida-lmstudio.shor run it from terminal
A public URL like https://xyz.trycloudflare.com will appear — copy it.
Step 6 — Connect AIDA to LM Studio
- Open AIDA → Settings
- Under LLM Provider, click LM Studio (Local)
- Paste the tunnel URL in the Server URL field
- Click Save URL
- Click Auto-Detect Models
- Select your loaded model from the list
- Start chatting!
Tools for Local Models
Local models run chat-only by default — enable tools with the Tools for local models setting on the LM Studio card in Settings (None / Curated / Full). Curated needs a 3B+ model; Full is for 8B+ models. If you enable tools, raise the context length in LM Studio's server settings to 8192 or more — the default is too small and causes ignored tools, cut-off answers, or 524 timeouts. Full details are in the Ollama doc's "Tools for Local Models" section — the tiers work identically here.
Tips
Start order matters — always start LM Studio server first, then run the launcher.
Keep both running — LM Studio server and the launcher must stay open while using AIDA with local models.
New URL each session — the tunnel URL changes every restart. Update it in AIDA Settings each time.
Slow responses? Use a smaller model (0.5B or 1.5B). A dedicated GPU speeds up inference significantly. Enable GPU acceleration in LM Studio → Settings → GPU.
524 timeout error — means the model took too long to respond. Switch to a smaller model or enable GPU acceleration.
Context window — lower the context window in LM Studio server settings for faster responses on slower hardware.