Keep thinking local.
Run supported MLX models on Apple silicon. Model inference happens on your Mac, without a paid AI-provider subscription.
ESSENTIAL EDITION 1.0 · FOR MAC
A space to think, write and explore with AI running on your own Apple-silicon Mac.
Meet HALRON. Codename MetalHead. Download supported models, work with your documents, and keep your conversations close.
Pay what you can. USD 20 suggested, from USD 0.
One-time purchase

LOCAL INTELLIGENCE. PERSONAL CONTROL.
Run supported MLX models on Apple silicon. Model inference happens on your Mac, without a paid AI-provider subscription.
Organize conversations in notebooks and nested folders. Work with text, searchable PDFs, and images supported by your chosen model.
Listen with the built-in Reader. Export sessions to Markdown, HTML or PDF, and back up your local work.
BEFORE YOU BEGIN
macOS 26 · M-series Mac · 8 GB memory minimum
Model requirements vary. Larger models need more memory and storage. Internet access is used for model downloads, licence activation and refresh, and optional news checks.
First launch may require approval in macOS Privacy & Security. See the installation steps →