Anthropic Launches Two New AI Models as It Lines Up Mega IPO

Having a fresh product to showcase is advantageous when seeking $100 billion from investors, as Anthropic is preparing to do.

Anthropic today introduced Claude Fable 5.2 and its equally unrestricted Mythos 5.2. According to the company, these models deliver another major performance gain at a reduced cost. The initial perception, however, is that the advance is not sizable.

The announcement comes at a strategic moment: Anthropic is getting ready for what might become the largest IPO in history, with a possible raise of $80 billion to $100 billion at a valuation of up to $1.75 trillion.

Anthropic asserts that the new versions beat Opus 5.5 on nearly every benchmark it evaluated, notably topping OpenAI's GPT-6 Astra in agentic coding. Fable 5.2 logged 69.9% on FrontierSWE v2, compared with Astra's 65.5%.

Testers have found that Fable 5.2 and Mythos 5.2 communicate more clearly than past Fable and Mythos models. They generate outputs up to 30% faster than Claude Fable 5.1 and Claude Mythos 5.1, and they’re available at a lower price: $8 per million input tokens and $40 per million output tokens.

That pricing is roughly 20% below earlier benchmarks.

The price reduction may be what investors find most significant about the announcement.

A central question facing the AI sector is whether its economics are fundamentally sound. While demand is undeniable, delivering ever more capable models remains extremely costly.

According to Anthropic, Fable 5.2 configured for High effort is able to equal Opus 5.5 set to Max effort when performing coding tasks, with costs around 50% less.

That is a worthwhile improvement and fits the trajectory of the AI boom, but it is not the kind of revolutionary development that would stir intense excitement for the IPO.

The company also states that Claude now generates over 90% of the code incorporated into its own main codebase. Such a figure is likely to be highlighted during the IPO marketing.

There are also indications that the economics are improving: computing expenses have gone from 172% of revenue to 57% over the past year. That still means a costly operation, but the decline is dramatic and hints at additional efficiencies ahead.