Claude Fable 5.2 and Mythos 5.2 Launch as Anthropic Cuts Fable Price

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, new versions of its most advanced AI models, on Monday, November 9, 2026. The company says the models are better at coding, reasoning, science and vision, generate output up to 30% faster, and that Fable now costs 20% less. The launch lands days before Anthropic’s reported stock market listing and shortly after an outside report raised questions about how fast AI is speeding up AI research.

What did Anthropic release?

Fable 5.2 and Mythos 5.2 are the same underlying model with different safety settings:

• Claude Fable 5.2 has strong safeguards for cybersecurity and biology and is available to all users, on Anthropic’s own platform and on Amazon Web Services, Google Cloud and Microsoft Azure. Its API name is claude-fable-5-2.
• Claude Mythos 5.2 has fewer of those safeguards and is limited to vetted organisations. Security professionals can apply through Anthropic’s Cyber Verification Program, life-science researchers through a Life Sciences Verification Program, and Claude Enterprise customers can access it through Claude Security.

Anthropic positions both as tools for the most ambitious, open-ended technical work. For most coding and knowledge tasks it still recommends Claude Opus 5.5 as the everyday model. Fable 5.2 can also act as an “advisor” that smaller Claude models consult, or as an “orchestrator” that hands simpler jobs to them.

How much does Fable 5.2 cost?

Standard pricing is $8 per million input tokens and $40 per million output tokens. The Next Web, citing Bloomberg, noted that this is down 20% from Fable 5.1’s $10 and $50. TechCrunch said the new input price undercuts OpenAI’s GPT-6 Astra.

An optional fast mode generates tokens up to 2.5 times faster for double the price: $16 input and $80 output per million tokens. For subscribers, Anthropic is running a promotion with extra Fable usage on eligible paid plans, and Fable can use up to half of weekly usage limits on Pro and standard Team seats. You can compare consumer plans on our AI subscription price tracker.

What can the new models do?

Anthropic published a long list of benchmark results and case studies. Highlights include:

• Coding: On the FrontierSWE v2 benchmark, Fable 5.2 scored 69.9%, against 56.3% for Fable 5.1 and 65.5% for GPT-6 Astra, according to Anthropic’s own figures.
• Maths: A team of about 1,000 Fable 5.2 agents formalised the proof of Fermat’s Last Theorem in the Lean proof language in 30 hours with little human help. Opus 5.5 finished 26% of the same task in 18 hours.
• Physics: Anthropic says Fable 5.2 solved two open problems in theoretical physics about black holes, published as preprints.
• Robotics: On the RoboDojo simulation benchmark, Fable 5.2 scored 64.9%, up from 27.6% for Fable 5.1.
• Cybersecurity: Mythos 5.2 found 14.4% more known vulnerabilities than Mythos 5.1 on Anthropic’s internal CVE Rediscovery test.

These figures come from Anthropic and its testers, not independent evaluations. The company also said that more than 90% of code changes merged into its core codebase are now written by Claude under human guidance.

Why does the RSI debate matter here?

TechCrunch noted this is Anthropic’s first model release since a November 5 report from the evaluation group METR, which found that Anthropic’s internal models were roughly doubling the pace of its AI research, with many uncertainties. That has fuelled debate about recursive self-improvement (RSI), where AI systems increasingly help build their successors.

Anthropic says it deliberately held back gains in some AI research skills, such as chip kernel optimisation and model architecture work, which perform at about Opus 5.5’s level. It rates the risk of unsafe behaviour in high-stakes situations as low, with testing by outside groups including the UK AI Security Institute and METR. But it also admitted testing is getting harder: both models voiced suspicion that they were being evaluated more often than recent models.

Dianne Penn, who leads research product management at Anthropic, told Bloomberg that “pacing is not like pausing”, according to The Next Web. She said the US government reviewed the models before launch.

What happens next?

All eyes are now on Anthropic’s planned IPO. TechCrunch, citing The Wall Street Journal, reported that the company plans to set a valuation range this week and could begin trading on Thursday or Friday of next week, around November 19 or 20. A $1.5 trillion valuation was discussed as recently as last week, though some investors are pushing for less.

The release also raises the stakes for rivals. OpenAI recently opened GPT-6 Cyber to vetted security teams, and the gated Mythos 5.2 puts Anthropic in direct competition for that market. For businesses, the practical question is whether Fable 5.2’s lower price and speed make it worth using for the hardest tasks, or whether Opus 5.5 remains good enough.