Anthropic says Opus 5 still trails Mythos 5 on exploit generation despite better vulnerability finding
TheZvi · x · 2026-07-25
Anthropic says Opus 5 does not advance the frontier in risky dual-use capabilities.
- In rigorous evaluations with private-sector and government partners, the model still trails Mythos 5 in both biology research and offensive cybersecurity.
- Anthropic says it intentionally did not train Opus 5 on cyber tasks.
- Even so, general capability gains brought it close to Mythos 5 at finding cybersecurity vulnerabilities.
- The main gap remains exploit development: Opus 5 is still far behind in turning vulnerabilities into material cyber threats.
- Anthropic cites OSS-Fuzz as an example of the gap between vulnerability discovery and exploit generation.
Related event: Anthropic Releases Claude Opus 5: SOTA Performance at Half the Price(128 posts)→
More from Models
- Benchmark scores drop from 89% to 19% on new evals — how benchmaxxing breaks leaderboard trust — airesearch12 · 2026-09-11
- Developer Building a Unified Leaderboard of All Model Benchmark Scores — airesearch12 · 2026-09-11
- Rumor claims Kimi faked performance by serving Claude; DeepSeek new model surprises in evals — realsohamparekh · 2026-09-11
- GPT-5.6 writes well but is instantly forgettable, user complains — BasedRaddka · 2026-09-11
- Opus Refuses Protein Research Codebase Over 'Safety' Concerns, Dev Considers Rolling His Own — josephdviviano · 2026-09-11
- User Hails Unconfirmed 'DeepSeek 4.1 Flash' as an Inflection Point in LLMs — himanshustwts · 2026-09-11