Anthropic is in discussions about a potential initial public offering that could happen as soon as the fourth quarter of 2025, according to a report from The Information published Thursday. Bankers competing to lead the offering have told the company it could raise as much as $60 billion, the report said.
The IPO news landed on the same day Anthropic confirmed it is preparing to release Claude Mythos, its next flagship AI model. An Anthropic spokesperson told Fortune that Mythos represents "a step change" in AI performance and is "the most capable we've built to date."
The statement followed a Fortune report on an unreleased Anthropic blog post that referenced Mythos and described the model as carrying "unprecedented cybersecurity risks." That framing is not unusual at this stage of development — companies routinely evaluate pre-release models for safety risks and adjust them before launch. The unpublished post also said Mythos is expensive to run and that select customers are currently in testing, though Fortune noted it is not yet clear whether that testing has begun.
A New Tier Above Opus
The same unreleased blog post introduced a model tier called Capybara, which Anthropic described as "larger and more intelligent than our Opus models — which were, until now, our most powerful." The post said Capybara "gets dramatically higher scores on tests of software coding, academic reasoning, and cybersecurity" compared to Claude Opus 4.6, Anthropic's current flagship. It is not yet clear whether Mythos and Capybara refer to the same underlying model or are distinct.
If Capybara is a separate tier above Opus, it would represent a meaningful restructuring of how Anthropic positions its model lineup. The current hierarchy — Haiku, Sonnet, Opus — has held for several release cycles. A tier above Opus, benchmarked against coding and academic reasoning specifically, would signal that Anthropic is targeting a different kind of customer than the general consumer market.
OpenAI's Competing Release
Anthropic is not alone in preparing a new flagship. The Information reported earlier this week that OpenAI is working to release its own new model, codenamed Spud, after completing pretraining. The timing puts both companies in a position where their next-generation models arrive in roughly the same window, intensifying a benchmark competition that is already being watched closely by enterprise buyers.
What the IPO Framing Means
The convergence of the IPO reports and the Mythos disclosure — intentional or not — is notable. A $60 billion raise would be one of the largest technology IPOs in recent memory, and it would require Anthropic to demonstrate that its model roadmap supports a valuation at that level. Mythos, and especially the Capybara tier above Opus, functions as evidence in that argument. The timing of the disclosure, ahead of a formal IPO announcement, suggests Anthropic is already managing the narrative around what its next model can do.
No IPO date has been set. Anthropic has not confirmed a formal timeline for the Mythos release.
Comments