Claude Opus 5 cuts defensive vulnerability-discovery flags to 13.9%
TheZvi · x · 2026-07-25
A chart compares classifier flag rates for vulnerable-code discovery across Claude Opus 4.8, Claude Fable 5, Claude Sonnet 5, and Claude Opus 5.
- For violative vulnerability discovery, Opus 5 is flagged 99.0% of the time, close to Fable 5’s 100%.
- For defensive vulnerability discovery, where low flag rates are preferred, Opus 5 drops to 13.9%, much lower than Fable 5’s 90.0%.
The post argues this means Opus 5 significantly reduces block rates for legitimate defensive work while still catching harmful use cases.
Related event: Claude Opus 5 Eases Source Code Vulnerability Discovery Restrictions(2 posts)→
More from Models
- Claude Opus 5 can misread a document despite knowing the underlying facts — teortaxesTex · 2026-07-25
- Early Opus 5 feedback says Claude’s writing is now “4o-level slop,” despite stronger intelligence — jdjohnson · 2026-07-25
- Users say Anthropic’s new model feels faster and stronger than Fable — emax · 2026-07-25
- Claude Opus 5 scores 30.2% on ARC-AGI-3 public demo environments — GregKamradt · 2026-07-25
- Release blog teaser shows a near-tie on FrontierCode agentic coding benchmark — hardmaru · 2026-07-25
- User Accuses Anthropic of Gaming ARC-AGI-3 by Training Specifically on Benchmark Patterns — VraserX · 2026-07-25