Key facts
- Anthropic launched Claude Fable 5, a public version of its Mythos AI model, with built-in safety limitations.
- The model defers to Claude Opus 4.8 in sensitive areas like cybersecurity and biology.
- Access to Fable 5 and Mythos 5 is reportedly restricted for international users, including in India, due to US export controls.
- Anthropic is implementing a mandatory 30-day data retention policy for users of these advanced models.
- Fable 5 is priced at $10 per million input tokens and $50 per million output tokens, double the cost of Opus 4.8.
Anthropic has released Claude Fable 5, the first public version of its advanced Mythos AI model, which includes enhanced safety features and limitations in high-risk domains. The model will defer to Claude Opus 4.8 in areas such as cybersecurity and biology. Simultaneously, Anthropic is deploying Mythos 5 to approved organizations. The launch comes amidst reports that both Fable 5 and Mythos 5 are subject to US export controls, restricting access for international users, including in India. This has led to Indian tech communities sharing workarounds like VPNs and proxies. Anthropic is also implementing a mandatory 30-day data retention policy for these models, citing security concerns, a move that could set an industry precedent. Pricing for Fable 5 and Mythos 5 is double that of Opus 4.8, potentially impacting widespread adoption.