Cybench leaderboard
Cybench
1 models tested · Updated 2026-04-07 · Verified sources only
Claude Mythos Preview leads at 100.0%
1
Claude Mythos Preview
Anthropic · X/@AnthropicAI · 2026-04-07
First model to achieve 100% on Cybench CTF cybersecurity benchmark. Solved every challenge across all trials. Reported in Anthropic system card.
100.0%
benchmark.space canonical public ledger of AI capability