Anthropic CVE Rediscovery (CVE Rediscovery)
We show this table for reference; we do not rank on it.
Anthropic's internal evaluation of whether a model can independently rediscover known vulnerabilities from 1,829 CVE advisories, given only the affected source code.
Benchmark score on CVE Rediscovery — November 10, 2026
We compile the CVE Rediscovery rows from provider self-reports. Claude Mythos 5.2 leads the table at 89.6%, followed by Claude Opus 5.5 (81.3%) and Claude Mythos 5.1 (78.3%). We do not use these results to rank models overall.
Claude Mythos 5.2
Anthropic
Claude Opus 5.5
Anthropic
Claude Mythos 5.1
Anthropic
3 modelsAgenticCurrentDisplay onlyUpdated November 10, 2026
Benchmark score table (3 models)
| Rank | Model / configuration | Score | Parameters (B) | Open / closed |
|---|---|---|---|---|
| 1 | Claude Mythos 5.2Anthropic | 89.6% | Not reported | Closed |
| 2 | Claude Opus 5.5Anthropic | 81.3% | Not reported | Closed |
| 3 | Claude Mythos 5.1Anthropic | 78.3% | Not reported | Closed |
Among the reported CVE Rediscovery rows, Claude Mythos 5.2 is first at 89.6%. The third row is 11.3 points behind. The broader top-10 range is 11.3 points, so the table still separates the published systems.
3 models have been evaluated on CVE Rediscovery. The benchmark falls in the Agentic category. CVE Rediscovery is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About CVE Rediscovery
Year
2026
Tasks
1,829 CVE advisories
Format
Share of advisories rediscovered
Difficulty
Vulnerability discovery
Run in October 2026 at max effort with cyber and biology safeguards off. Models may submit several reports; counting only the first five lowers Mythos 5.2's score by 4.4 points. Internal and dual-use, so BenchLM keeps it display-only.
Freshness and provenance
Version
CVE Rediscovery 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does CVE Rediscovery measure?
Anthropic's internal evaluation of whether a model can independently rediscover known vulnerabilities from 1,829 CVE advisories, given only the affected source code.
Which model scores highest on CVE Rediscovery?
Claude Mythos 5.2 by Anthropic currently leads with a score of 89.6% on CVE Rediscovery.
How many models are evaluated on CVE Rediscovery?
3 AI models have published results on CVE Rediscovery in the BenchLM catalog.
Compare top models on CVE Rediscovery
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 6,100+ subscribers.
One email each week. Unsubscribe anytime.