Anthropic (opens in new tab) on Monday released Claude Fable 5.2 and Claude Mythos 5.2, new versions of its most advanced AI models. The company said the two share the same underlying model: Fable carries safeguards against cybersecurity and biology misuse and is available to all users, while Mythos is limited to vetted organizations in its Life Sciences and Cyber Verification programs.
Both models are priced at $8 per million input tokens and $40 per million output tokens. CNBC (opens in new tab) reported that is 20% less than the 5.1 generation, and TechCrunch (opens in new tab) reported that Fable's input price fell from $10. Anthropic (opens in new tab) said the new versions produce output up to 30% faster than their predecessors and offered an optional fast mode at twice the price.
Anthropic (opens in new tab) said that during pre-release testing the models solved two open problems in the physics of black holes, designed three working electronic devices and, running as a team of about 1,000 agents, formalized the proof of Fermat's Last Theorem in 30 hours. Those results and the benchmark scores the company published come from Anthropic (opens in new tab) and its testers and have not been independently verified.
On safety, Anthropic (opens in new tab) said it judges the risk of the models behaving unsafely in high-stakes situations to be low, but it also flagged a testing problem. "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 wrote. TechCrunch (opens in new tab) noted the release is Anthropic (opens in new tab)'s first since an outside report from the evaluation group METR suggested its internal models were speeding up the company's AI progress.
CNBC (opens in new tab) said the launch is Anthropic (opens in new tab)'s first frontier model release since chief executive Dario Amodei called in September for slowing the pace of AI development. "We called for pacing, not pausing," Dianne Penn, Anthropic (opens in new tab)'s head of product management for research, told the network. CNBC (opens in new tab) also reported the company is preparing for an initial public offering that could come as soon as next week.