Open Source — Apache 2.0
VEXT Pentest-7B
The first open-source AI model purpose-built for cybersecurity.
Trained on 260K+ curated security examples by VEXT Labs, Inc.
Category Breakdown
Pentest-7B vs General LLMs
Benchmark Overview
Key Highlights
- 100% on compliance benchmarks — SOC 2, PCI DSS, HIPAA, NIST CSF, ISO 27001
- 90% on practical pentesting — detects, exploits, and remediates real vulnerabilities
- 80% on web application and API security — core pentesting domains
- 2x better than general-purpose LLMs on security-specific tasks
- 7B parameters — runs on a single GPU or even CPU, deployable anywhere including air-gapped environments
- Includes opt-in community telemetry for continuous model improvement