Lightweight
Lamb
Small model. Big privacy.
- Fast & efficient
- Runs on modest hardware
- Great for edge & offline
- Yours to keep
Private. Local-first. Owned intelligence.
Not another cloud subscription.
Buy once. Run it locally. Own it forever.
Lightweight
Small model. Big privacy.
Orchestration
Coordinate. Orchestrate. Observe.
Flagship
Our most capable local models.
The independent intelligence workshop
Useful intelligence should feel like a well-made tool: close at hand, understandable and entirely yours.
Read the field notes →Inference stays on hardware you choose, whether that is a workstation, a private rack or an edge device.
Prompts, context and working memory remain inside your environment instead of becoming somebody else’s dataset.
Every model family is designed for useful capability per watt, per gigabyte and per machine you already own.
The model remains usable on your terms. No disappearing subscription, forced migration or rented access layer.
Recent research & notes