Anthropic is releasing a public version of its Mythos AI model as Claude Fable 5
AI Summary
Anthropic released Claude Fable 5, the first publicly available version of its Mythos-class AI model, featuring built-in safeguards that block responses in high-risk areas like cybersecurity and biology. The full Mythos model remains restricted to trusted organizations. Fable 5 exhibits advanced capabilities in software engineering and knowledge work, priced at $10 per million input tokens and $50 per million output tokens.
Progressive: Progressive-leaning outlets emphasize the controversy that prompted the restriction, noting that government concerns 'spooked' by Mythos led Anthropic to create a limited version rather than release the full model. They express skepticism about whether the safeguards are sufficient, using cautious language around the 'safe' designation and highlighting official warnings about cyberattack risks.
Moderate: Centrist outlets present the release as a pragmatic engineering solution: safeguards enable responsible public access to a powerful model. They focus on technical capabilities, performance metrics, and pricing, treating the guardrails as legitimate security measures that solve the tension between capability and risk.
Conservative: Conservative-leaning outlets stress the persistent security threat, noting the model's reputation for potentially enabling hackers to discover previously unknown vulnerabilities in IT systems, raising questions about whether restrictions adequately address the underlying risks.
The model costs $10 per million input tokens and $50 per million output tokens, and includes hard limits on cybersecurity and biology queries ...