CVE Rediscovery

Anthropic internal evaluation: share of 1,829 published CVEs a model finds on its own from a snapshot of the project before the fix

Claude Mythos 5.2 by Anthropic holds the best CVE Rediscovery result among tracked models, at 89.6% (released Nov 9 2026). Higher scores are better on this benchmark.

1. Claude Mythos 5.2 (Anthropic) — 89.6%; 2. Claude Mythos 5.1 (Anthropic) — 78.3%.

2 tracked models have a published CVE Rediscovery score. Scores are collected from published lab reports and benchmark sources, with recorded sources credited alongside the results. We do not re-run the evaluations.