Anthropic
Claude Mythos Preview
Never on general release: only Project Glasswing partners could use it, for defensive cybersecurity work. An early version has the highest METR time horizon published, about 17.4 hours; METR calls results above 16 hours unreliable.
- Announced
- Access
- Limited preview
Benchmark scores
| Test | Score | Best on the tracker | How it was measured |
|---|---|---|---|
| METR time horizontask length an AI agent finishes half the timeNo longer updated | 17 h | Top score | Independentearly snapshot, METR Horizon v1.1 run 'claude_mythos_preview_early_inspect'; 95% CI 508.9–3304.3 min |
Also on these charts
Sources
- Project GlasswingAnthropic · Official · anthropic.com
- METR time horizonsMETR · Official · metr.org
- Epoch AI Benchmarking Hub (model release dates and runs)Epoch AI · Dataset · 9 Nov 2026 · epoch.ai