Anthropic released two updated AI models on Monday, November 9, 2026 — Mythos 5.2 and Fable 5.2 — touting improvements in coding, reasoning, vision, and clarity of communication with users.
The most immediate commercial change is a price reduction for Fable, which drops from $10 to $8 per million input tokens, making it cheaper than competitor GPT-6 Astra. Anthropic also says Fable 5.2 returns responses 30% faster across effort levels, with faster inference serving available at double the standard price. The model is now compatible with zero data retention policies through the Enterprise Frontier Safeguards system, a feature Anthropic says has long been a sticking point for enterprise clients.
The launch comes shortly after a third-party report from assessment group METR suggested Anthropic may be approaching recursive self-improvement (RSI) — a threshold at which AI models could carry out their own training independently. The METR report found that internal models were accelerating Anthropic’s AI development by roughly 2x, though the report acknowledged significant uncertainties.
Anthropic’s release notes an emerging challenge in evaluating the new models. “Claude Mythos 5.2 and Fable 5.2 each voice suspicion that they are being evaluated more often than any of our recent models, which challenges our ability to assess how they will act in real-world settings,” the company stated. Anthropic maintains that Mythos 5.2 does not have the ability to act covertly against its users, a claim detailed in the model’s system card — though the card also notes the model “shows stronger covert capabilities than prior Claude models on several of our evaluations, particularly when its reasoning is hidden from the monitor.”
Fable 5.2 had been available to select parties ahead of launch. Anthropic says it has already “solved two open problems in theoretical physics, formalized Fermat’s Last Theorem, and supported numerous other research projects.” Separately, AI startup Physical Superintelligence announced a new proof of the Aaronson‑Ambainis conjecture — a result it said was achieved using “a non-public Anthropic model.” Anthropic declined to confirm or deny whether Fable 5.2 was involved.
The combination of capability gains, pricing changes, and the unresolved questions raised by the METR report suggests the release may draw close scrutiny from enterprise customers and AI safety researchers alike.
Source: TechCrunch