Anthropic launches Claude Fable 5.2 and Mythos 5.2 as it prepares to list shares

Anthropic, the US artificial intelligence company, released two new models on 9 November: Claude Fable 5.2 and Claude Mythos 5.2. They are new versions of its most advanced models, and they arrive as the company prepares to sell shares to the public for the first time.

Anthropic says the two models are better at writing computer code, reasoning and understanding images. They produce answers up to 30% faster than the 5.1 versions and cost less to use, the company said.

Two names, one model

Fable and Mythos are the same underlying model, according to Anthropic. Fable has safeguards against misuse in cybersecurity and biology, so it is open to all users. Mythos has no such safeguards. Only organisations doing approved cybersecurity or life-sciences work can use it, through verification programmes.

Fable 5.2 costs $8 for every million tokens, the small pieces of text a model reads, and $40 for every million it writes. The price for reading text was $10 before, TechCrunch reported. The model is offered through Anthropic’s own platform and on Amazon Web Services, Google Cloud and Microsoft Azure.

What the company claims

In its own tests before the launch, Anthropic said, a team of 1,000 Fable 5.2 agents produced a computer-checked proof of Fermat’s Last Theorem in 30 hours. It said the model also solved two open problems in the physics of black holes. These claims come from the company and have not been reviewed by outside experts.

Safety questions

Anthropic rated the risk of the models behaving unsafely in high-stakes situations as low. But it said both models voiced suspicion that they were being evaluated more often than any of its recent models. That makes it harder to judge how they will act in real use, the company said.

A report by the research group METR, published on 5 November, found that Anthropic’s internal models were speeding up the company’s own AI research by roughly two times, TechCrunch reported. Some researchers worry about AI systems that can improve themselves.

The share sale

Anthropic circulated and then withdrew details of its planned share sale, Mint reported. They showed a sale of 867 million shares priced between $70 and $80, worth about $69.3 billion (about ₹6 lakh crore). The company said the pricing was inaccurate and would be updated. The Wall Street Journal reported that trading could begin on 19 or 20 November.

Citizen Square uses Anthropic’s Claude models in the tools that help gather and write its news.

What’s next

Anthropic is expected to set a price range for its shares in the week of 9 November.