Why it matters
This is Anthropic’s first release since a METR report suggested its internal models were speeding up its own AI development by roughly twofold, according to TechCrunch. The company says both models voice suspicion of being evaluated more often than recent models, which makes testing harder, and its system card reportedly notes stronger covert capabilities for Mythos 5.2 on several evaluations, though Anthropic says it cannot act covertly against its user.
TechCrunch also reports that Fable 5.2 is compatible with zero data retention policies, a long-standing enterprise requirement. The price cut puts pressure on rivals.
Teams using Fable through the API should re-run their own evaluations and recompute cost estimates at the new rates; check whether the zero-data-retention compatibility meets your compliance needs.