Anthropic has released Claude Fable 5, a publicly accessible version of its advanced Mythos model, featuring enhanced capabilities in software engineering, knowledge work, and vision. The model includes stringent safety guardrails, deferring to Claude Opus 4.8 for high-risk domains like cybersecurity and chemistry. This launch follows Anthropic's recent warnings about the rapid advancement of AI and the potential for recursive self-improvement, while also introducing a mandatory 30-day data retention policy for enhanced security monitoring. AI
IMPACT Sets new SOTA on complex analytical tasks and coding; pressures competitors with advanced safety features and higher pricing.
RANK_REASON Frontier-lab model release with system card [lever_c_demoted from frontier_release: ic=2 ai=1.0]
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →