Documentation
Everything you need to start
The fastest path is the desktop app. The guide below covers every screen, step by step.
Quick start
Install the desktop app
Download the installer for Linux. The compiler is embedded: nothing else to set up. macOS and Windows are coming soon.
DownloadStart from a published model
Open one of the reference architectures, or paste a Hugging Face model ID or config.json. It is read locally.
ReadConnect an AI provider
Bring your own key for OpenAI, Anthropic, Gemini, Mistral, Fireworks, DeepSeek or GLM, or point the assistant at any custom endpoint.
ReadSecurity
What leaves your machine (nothing), how keys are handled, and what the assistant can and can't see.
Read
Browse the guide
The same guide that is built into the desktop app, with its screenshots. In English for now.
- Getting started4 pagesWhat NEURAX does for youWhat NEURAX does not do yetYour first analysis+ 1 moreOpen
- Step by step7 pagesAdd your datasetTell NEURAX your machineStart from a template+ 4 moreOpen
- Your machine and your data2 pagesWhat NEURAX knows about this computerPointing at your dataOpen
- Building a model5 pagesWorking on the canvasRepeated layersArchitecture families+ 2 moreOpen
- Bringing models in3 pagesImporting a HuggingFace config.jsonImporting a ModelScope config.jsonOpening a saved designOpen
- Saving and sharing3 pagesJSON and TOML design filesProjectsExportingOpen
- Working day to day3 pagesUndo and redoNeurax AIKeyboard shortcutsOpen
- The workspaces1 pageWhat each one showsOpen
- Reading the results3 pagesWhat the numbers meanDiagnostics — where the optimisation isAccuracy and limitsOpen
- Troubleshooting2 pagesWhen something looks wrongWhere NEURAX keeps thingsOpen