Anthropic Releases Claude Fable 5.2 and Mythos 5.2

Anthropic has released two new AI models, Claude Fable 5.2 and Claude Mythos 5.2, available from November 9 through cloud platforms AWS, Google Cloud and Azure. The company says the models answer up to 30% faster than their previous versions, Fable 5.1 and Mythos 5.1, and reach the performance of Anthropic's Opus 5.5 model for roughly half the cost per task. Mythos 5.2, a version built for security work, scored 89.6% at finding known software vulnerabilities in a company test, up from 78.3% for Mythos 5.1.

For companies already building products on Claude, this points to lower token bills without giving up quality, and room to hand over heavier agentic work, tasks where the model acts on its own over many steps, such as code review and vulnerability hunting. The gains show up mainly in technical benchmarks rather than everyday conversation, which signals where competition among AI labs is heading: toward models that can carry out long technical jobs by themselves, not just answer questions.