About Aitium
We build the models. You run them.
Aitium builds small specialized models, the engine that serves them on one or two 24GB GPUs, and the runtime that keeps AI agents inside the boundaries their operators set. We publish what we measure, including what fails.
Aitium takes its name from the Artificial Intelligence Consortium, the group the company grew out of.
Mission
What we are for
Useful AI that runs on hardware you control, with limits you can enforce and results you can check.
Proof
Check the work
Where the code, weights, results and engineering notes are published.
- Evaluation resultsCyberGuard results, including the misses.Coming soon
- Engineering notesWhat broke, why it mattered, and the fix.
What we build
Models and the platform that runs them
CyberGuard
A small specialized model that sorts security findings and reasons about threats, on one 24GB GPU inside your own perimeter.
Explore CyberGuardCogenics
Matches DNA variants to medicine compatibility, grounded in CPIC prescribing guidelines and ClinVar variant data.
Explore CogenicsFoxtail Harness
Foxtail Harness is the runtime between AI agents and production systems. Scopes, spend caps, review gates, and a full audit trail.
Explore Foxtail HarnessEmber
Serves small models across 1–2 24GB GPUs through an OpenAI-compatible API — private inference on hardware you control.
Explore Ember