Detect malicious code in AI models before deployment
No signup • Free API • Real-time results
| Feature | Scanner Free | Inventory Open Source | Offensive Enterprise |
|---|---|---|---|
| Scan models | ✓ Unlimited | — | ✓ |
| API access | ✓ Free key | — | ✓ |
| Asset tracking | — | ✓ Self-host | ✓ |
| Red teaming | — | — | ✓ |
| SOC integration | — | — | ✓ |
| Support | Community | Community | Priority |
| Pricing | $0 | GitHub | Contact |
| Contact |
Hidden code that executes when models load
Runs unauthorized commands on your system
Malicious objects created during loading
Hijacks your GPU to mine cryptocurrency
Steals your data and sends it to attackers
Hidden switches that activate malware later
Manipulated weights alter model behavior
Disguised code designed to evade detection
Scanned 50,000+ models in production
8 detection engines, not just signatures
Most models scanned in under 1 second
Core scanner releasing Q2 2026