Anthropic releases Claude Fable 5.1, with a restricted twin for science and cyber

Anthropic has released Claude Fable 5.1, a new version of its AI system designed for difficult coding, research, and office work that may take hours and require many steps. The important wrinkle is that Claude Mythos 5.1 is not a separate brain: it is the same underlying model, but available only to vetted cybersecurity and life-sciences organizations with fewer restrictions in those sensitive fields. Anthropic says its tests show gains in tasks such as running scientific investigations and molecular design, an encouraging but still company-reported sign that AI is moving from answering questions toward helping researchers explore and check ideas. For ordinary customers, Fable is the version on offer; Anthropic is also cutting the cost of reusing previously processed material, which matters most when software agents repeatedly consult large files.