Anthropic Releases Fable 5.2 and Mythos 5.2, 20% Cheaper and 30% Faster, Its First Frontier Models Since Amodei's Slowdown Call

Since Anthropic filed to list on the Nasdaq as ANT and the tech selloff pulled valuation expectations down, the company has put product on the table. Today, Nov. 9, it released Claude Fable 5.2 and Claude Mythos 5.2, the first frontier models it has shipped since Dario Amodei urged the industry in September to slow the pace of AI capability gains.

What Anthropic shipped

Anthropic says both models generate output 30% faster than the 5.1 generation. Pricing drops 20%, to $8 per million input tokens and $40 per million output tokens, from $10 and $50.

Fable 5.2 is generally available today on Pro, Max, Team and Enterprise plans. It is also available through the API and on Amazon Web Services, Google Cloud and Microsoft Azure. An optional fast mode runs up to 2.5 times normal speed at $16 and $80 per million tokens.

Mythos 5.2 is not for the general public. It goes only to vetted organizations in Anthropic's Life Sciences and Cyber Verification programs.

According to Anthropic's announcement, the two are one model with different safeguards. Fable carries classifiers that block high-risk cyber work such as exploit generation and binary vulnerability analysis, plus much biology research. Blocked requests are answered by an older Opus model instead. Mythos removes those restrictions for verified users.

Anthropic said the models went through rigorous safety and alignment testing, and that it built new evaluations and post-launch monitoring tools for the release.

'Pacing, not pausing'

Dianne Penn, Anthropic's head of product management for research, said the company's position has been misread. "We called for pacing, not pausing," she told CNBC.

Amodei's essay, "We Must Pace the Frontier," does say that. It also says plainly: "We must slow the pace at which we improve the capabilities of AI models." The same essay says pacing "does not mean halting model training or technical progress."

Penn said Fable 5.2 is meant for "power users or power use cases," meaning hard problems in engineering, mathematics and robotics. Opus 5.5, released in September, remains Anthropic's recommendation for everyday work.

The acceleration question

The release landed days after two documents that make the pacing argument harder to wave away.

On Nov. 5, the Berkeley nonprofit METR, which Amodei's essay named as a possible embedded outside evaluator, published an assessment based on about 11 weeks of internal data and staff interviews. METR wrote that "Anthropic appears to have made as much progress on benchmarks in the past year as they did between October 2023 and October 2025." It estimated Anthropic is improving its models on well-specified tasks about twice as fast as a year earlier.

METR attached caveats. It said it could be overstating the acceleration, which might be confined to well-specified tasks. It also said Anthropic's redactions limited what it could report.

Then there is Anthropic's own Nov. 2 amended IPO filing. It says the company's internal estimates for future models "indicate that we may imminently meet our Responsible Scaling Policy's definition of 'dramatically accelerated progress attributable to AI.'" No current model has been designated as exceeding that threshold. Anthropic said it is considering updates to the Responsible Scaling Policy.

Anthropic's answer is the testing and monitoring it describes. METR's assessment and the filing do not say the new models fail that testing. They do put numbers on how fast the company is moving.

The IPO pricing

The launch comes as the company tries to hold its listing window. CNBC reported that Anthropic is aiming for a valuation of $1 trillion to $1.5 trillion, down from a $2 trillion target floated in recent months, and that the IPO could come as soon as next week.

The Financial Times, citing multiple investors in the company, reported a somewhat different range: raising $80 billion to $100 billion at a valuation of $1.25 trillion to $1.75 trillion, before Thanksgiving. That compares with a current private valuation of $965 billion. One large investor told the FT, "My view is they will price it very conservatively, call it $1.5tn, and go up from there." Others expected roughly $1.3 trillion.

Those are investor and press estimates, not Anthropic figures. The company declined to comment on the FT report.

The backdrop is rough. The FT noted the Nasdaq 100 fell 6.4% on Thursday on worries about a slowdown in AI infrastructure buildout. Bloomberg reported that OpenAI, which delayed its own IPO plans, closed a private round of roughly $11 billion at a $1.25 trillion valuation. OpenAI had sought $30 billion or more at $1.4 trillion.

People familiar with the plans told the FT that Anthropic holds $150 billion in cash and liquid investments and expects to raise enough that it will not need to sell equity again.

What comes next

The FT said JPMorgan, which leads U.S. retail sales of the shares, is expected to hold an information session for prospective investors on Nov. 13. Anthropic's executives still have to run the roadshow and persuade buyers that shipping faster, cheaper frontier models is consistent with the pacing they asked the industry to adopt.

Whether the company will formally update its Responsible Scaling Policy before pricing is not known. Anthropic has said only that it is considering changes.