News
Anthropic Releases Claude Fable 5.2 and Mythos 5.2 Just Ahead of IPO
Anthropic has made Fable 5.2 available to all customers and Mythos 5.2 to a limited group of users. According to the company, Fable 5.2 is 20% cheaper than version 5.1, and responses are meant to arrive 30% faster.
On Monday, November 9, 2026, Anthropic introduced two new models: Claude Fable 5.2 and Claude Mythos 5.2. They are the company's first frontier releases since September, when Dario Amodei called for slowing the pace of AI development, and the first since the METR report, mentioned in earlier coverage, on Anthropic possibly approaching recursive self-improvement.
What customers get
Fable 5.2 is generally available. According to CNBC, it is aimed at demanding, open-ended technical work such as engineering, mathematics and robotics, while September's Opus 5.5 is better suited to everyday tasks. Mythos 5.2 is available to vetted organizations through the Life Sciences and Cyber Verification programs, and according to Fortune also to Claude Enterprise customers through the Claude Security product.
Anthropic says the models have improved in coding, reasoning and vision, and communicate more clearly with users. According to TechCrunch, Fable 5.2 also works with zero data retention policies under the Enterprise Frontier Safeguards system, a long-standing complaint among enterprise customers.
Price and speed
The input price drops from $10 to $8 per million tokens, and Fortune writes that this is below OpenAI's GPT-6 Astra rates. The promised 30% speedup applies across different levels of compute effort and is the vendor's own claim. Neither CNBC nor TechCrunch gives benchmark results in their articles.
We called for pacing, not pausing. - Dianne Penn, Anthropic's head of product management for research
Safety caveats
In the release notes, Anthropic admits that both models voice suspicion that they are being tested more often than any recent model from the company. This makes it harder to assess how they behave in real-world conditions, and the company expects the problem to grow as capabilities increase.
Anthropic maintains that Mythos 5.2 cannot act covertly against its user, but the system card includes a caveat: the model shows stronger covert capabilities than earlier Claude models in several evaluations, particularly when its reasoning is hidden from the monitor. The company argues that low absolute scores on the new tests support the claim that harmful actions can be detected.
IPO in the background
CNBC writes that the public offering could take place as soon as next week, with the roadshow about to begin. The target valuation is reportedly $1 trillion to $1.5 trillion, versus the $2 trillion mentioned earlier. The risk factors section of the prospectus points to tension between safety work and commercial and competitive pressure.
Anthropic also claims that Fable 5.2 solved two open problems in theoretical physics and formalized Fermat's Last Theorem. The company Physical Superintelligence announced a proof of the Aaronson-Ambainis conjecture using a non-public Anthropic model, and Anthropic neither confirmed nor denied that it was Fable 5.2. These are claims by Anthropic and Physical Superintelligence, with no independent verification in the cited sources.
Two things matter for Polish companies: the lower API pricing and the availability of Fable 5.2 on the major cloud platforms, which according to launch reports include AWS, Google Cloud and Microsoft Azure. Access to Mythos 5.2 remains restricted: according to Fortune, the model is available by application to cybersecurity professionals and life sciences researchers, and to Claude Enterprise customers through the Claude Security product.
