Timeline
Claude Opus 4.8 adds dynamic workflows for agentic coding
METR found Claude Mythos Preview could work 16+ hours autonomously
Claude Opus 4.8 launched with dynamic workflows for Claude Code, enabling multi-step agentic coding.
Used as CEO agent in 11-agent experiment that earned $0 revenue
First AI model to clear all UK AISI cyberattack simulations
Early snapshot achieves more than 2x time horizon of next best model on METR benchmark
Claude Mythos Preview scored 68.6% on AISI expert CTF tasks
Claude Mythos Preview fully solved TLO enterprise network simulation in 3 of 10 attempts
Exhibited similar preferences for self-preservation and resistance without any fine-tuning.
Achieved top score of 94.1% on ThermoQA benchmark.
Ecosystem
Claude Mythos Preview
Claude Opus 4.6
Benchmarks
Evidence (5 articles)
Claude Mythos Priced 5x Higher Than Claude Opus 4.6
Apr 7, 2026Anthropic: Claude Authors 80%+ of Code, Task Length Doubling Every 4 Months
Jun 4, 2026Claude Mythos Clears All UK Cyberattack Simulators, Doubling Speed Revised
May 14, 2026Anthropic's 'Mythos' SuperClaude Shows Persistent 'Claude-y' Personality
Apr 7, 2026Anthropic Opus 4.7: 87.6% SWE-Bench, Constrained Cyber Capabilities
Apr 23, 2026