Anthropic launches Claude Mythos 5.2 for vulnerability research
Original title: Anthropic Launches Claude Mythos 5.2 With 76.3% Cybersecurity Benchmark Score to Find 0-Day Vulnerabilities
Anthropic reports 76.3% on CyScenarioBench and 89.6% rediscovering 1,829 known CVEs, not unknown zero-days.
Anthropic released Claude Mythos 5.2 on November 9 alongside Claude Fable 5.2 for finding, reproducing, and fixing software flaws. Access is controlled through a Cyber Verification Program, while enterprise customers can use Mythos 5.2 through Claude Security without joining it. Anthropic says the models share one underlying model but differ in safeguards, and both are up to 30% faster than their predecessors. The company says the launch does not establish a 76.3% zero-day detection rate.
Unconfirmed: Cross-lab CyScenarioBench comparisons are limited, and outside tester results are not broad proof.
- Mythos 5.1 scored 61.7% on CyScenarioBench, 14.6 points lower.
- Opus 5.5 without safeguards scored 81.3% on the October 2026 CVE test.
- Standard pricing is $8 per million input tokens and $40 per million output tokens.
- Red Team Access covers authorized penetration testing and exploit development.
- Claude Security requires human review and approval of proposed patches.
- Vendors
- Anthropic
- Products
- Claude Security
From the original article
2 min read at cybersecuritynews.com
Anthropic has launched Claude Mythos 5. 2, its latest AI model for cybersecurity engineering, reporting a 76. 3% score on CyScenarioBench.
By Kavichselvan · cybersecuritynews.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.