Anthropic Releases Claude Fable 5.2 and Mythos 5.2, Cutting Prices by 20% Ahead of IPO

The launch matters for two reasons. It is the first frontier release since chief executive Dario Amodei called in September for the industry to slow the pace of AI development, and it comes days before what could be one of the largest stock market listings on record. CNBC reported that Anthropic executives are about to start the roadshow for an IPO that could take place as soon as next week.

The models are also cheaper. According to CNBC, the 5.2 generation costs 20% less than 5.1, at $8 per million input tokens and $40 per million output tokens.

Anthropic is handling the two models differently. Fable 5.2 is generally available from Monday. Mythos 5.2 will at first go only to vetted organisations through the company's Life Sciences and Cyber Verification Programs. Anthropic first introduced Mythos in April and restricted its rollout because of its advanced cybersecurity capabilities. Fable, a publicly available version of Mythos with additional safeguards, followed about two months later.

The company said Fable 5.2 handles difficult software engineering work, can coordinate teams of agents and can run long projects. Anthropic said that in pre-release testing the two models "helped our engineers complete otherwise-infeasible software migrations, designed three working electronic devices from scratch, solved two open problems in the physics of black holes, made substantial progress on robotics evaluations, and formalized the proof of Fermat's Last Theorem in 30 hours." No benchmark scores were cited in the coverage of the launch.

Dianne Penn, Anthropic's head of product management for research, told CNBC that Fable 5.2 is aimed mainly at "power users or power use cases", meaning complex and open-ended technical problems in fields such as engineering, mathematics and robotics. She said Claude Opus 5.5, released in September, remains better suited to everyday tasks, and that frontier capabilities could later move down the range: "We expect that over time to maybe trickle to the Opus class."

Penn also addressed the apparent tension with Amodei's September remarks. "We called for pacing, not pausing," she said. The company said the models went through extensive safety and alignment testing, and that it built new evaluations and post-launch monitoring tools before the release.

The timing gives investors a fresh look at Anthropic's product pipeline as they weigh the offering. CNBC reported that the company is aiming for a valuation of $1 trillion to $1.5 trillion, down from a target of roughly $2 trillion discussed in recent months. Anthropic filed its prospectus in late October. The risk-factors section describes the pull between its safety work and commercial and competitive pressure. In an amended filing, the company reported revenue of $33.24 billion for the nine months to September 30, roughly 1,200% higher than in the same period a year earlier.

Outside estimates vary widely. Morningstar analyst Malik Ahmed Khan on Monday put a $1.7 trillion fair value on Anthropic, with a bear case of $500 billion.

For enterprise buyers, the immediate changes are lower prices and faster output on Anthropic's top generally available model. Customers in regulated security and life-sciences work will need to go through Anthropic's verification programmes to get Mythos 5.2.