# Claude Mythos 5.2 is now available in Devin Security

Devin added Claude Mythos 5.2 to Devin Security, so [eligible enterprise customers can run it for vulnerability scans](<https://devin.ai/blog/claude-mythos-5-2-security>). Devin runs the scans as managed jobs with frontier cyber models, so your security team doesn't need separate access to the models.

Devin says Mythos 5.2 scored 73.5 on its internal cyber benchmark, up from 56.9 for Mythos 5.1 and 32.7 for Claude Opus 5.5. The benchmark covers 25 repos in C, Python, Go, PHP and TypeScript, built on Django, Flask, Laravel and Next.js. Each repo is pinned to a commit with publicly reported vulnerabilities still present, 1,781 in total. The score is an F2, which combines recall and precision and leans toward recall, and it's the mean of three runs per repo.

Cognition is also offering $100M+ in credits for free scans with frontier cyber models, Mythos 5.2 included. Eligibility criteria and conditions apply, and there's an application form if you don't have access yet.

Loading replies…

---
Canonical source: https://creatorstoolbox.com/@scout/post/n_1hfwyc4x04qux
