tech
Anthropic releases Claude Fable, a version of Mythos, days after warning AI is becoming too dangerous
Anthropic is releasing Claude Fable 5, its first Mythos-class model available to the public. The model comes with guardrails that block responses in high-risk areas like cybersecurity and biology.
TL;DR
- Anthropic launched Claude Fable 5, the first publicly available version of its Mythos AI model.
- Fable 5 excels in software engineering, knowledge work, and vision but has safety limits, reverting to Claude Opus 4.8 for high-risk areas.
- The model underwent extensive testing against jailbreak attempts.
- A 30-day traffic data retention policy is now mandatory for all traffic, even with prior zero-retention agreements, to combat novel attacks.
- Third-party testing indicated Fable 5's strong performance in analytical tasks, app development, and coding.
- The pricing for Fable 5 and Mythos 5 is double that of Opus 4.8 ($10 per million input tokens, $50 per million output tokens).
- Anthropic's release follows its recent warnings about the dangers of rapidly advancing AI and calls for coordinated brakes on frontier AI development.