Anthropic Launches Claude Fable 5: Mythos-Class AI Public

Anthropic has launched Claude Fable 5, its most advanced artificial intelligence model. This is the first publicly available model in its new Mythos-class family. What's interesting is that this technology was previously only available through limited testing. The launch follows the introduction of Claude Mythos Preview, which showed an unexpected ability to identify vulnerabilities in operating systems and web browsers. To address misuse concerns, Anthropic has built safeguards into Claude Fable 5. These restrict responses in high-risk areas like cybersecurity and biology. Users seeking information in these domains will be directed to the Claude Opus 4.8 model instead. Alongside this launch, Anthropic highlighted progress in Project Glasswing, a cybersecurity initiative. Participants include Amazon Web Services, Apple, Google, Cisco, Microsoft, and JPMorgan Chase. These organizations received early access to Mythos Preview to identify vulnerabilities. Anthropic says extensive testing confirmed researchers couldn't consistently bypass the model's safety controls. The company also announced that organizations with previous Mythos Preview access will be upgraded to Mythos 5. Both Claude Fable 5 and Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens. The bottom line is this launch signals Anthropic's continued growth in the competitive AI market.