Docs/Model Setup

Model Setup

Download and configure local AI models directly in NeuroTerm. No external tools required. 100% offline.

How It Works

NeuroTerm runs AI models locally on your machine using native GGUF support. Models power:

  • Neuro Input — Natural language to terminal commands
  • Local RAG — Ask questions about your datasheets
  • Neuro Tools — Issue reports, summaries, and more

1. Open Settings

1

Click the gear icon

Open Settings from the status bar or use Ctrl+,

2

Go to Neuro tab

Find the AI model configuration section

3

Click Download Model

Choose from recommended models or enter a custom URL

2. Choose a Model

Model Categories

  • Balanced — Good speed and quality tradeoff (~2GB)
  • Deep — Maximum quality, slower responses (~4GB)

Select a category and NeuroTerm downloads the right model. Models are stored locally and only need to be downloaded once.

3. GPU Acceleration (Optional)

NeuroTerm automatically uses your GPU if available:

NVIDIA GPU
Automatic CUDA acceleration
CPU Only
Works on any machine, just slower

No configuration needed. NeuroTerm detects your hardware automatically.

Troubleshooting

Download failed

Check your internet connection and try again. You can also download GGUF models manually and point to the file in Settings.

Slow responses

Try a smaller model. If you have an NVIDIA GPU with 4GB+ VRAM, larger models will run faster.

Out of memory

Switch to a smaller model. 3B parameter models typically need 4GB RAM, 7B models need 8GB+.

Next Steps