Capybara leak raises questions about Anthropic's IPO timing

An accidental leak revealed Anthropic's Capybara model while Bloomberg reported the company is eyeing an October IPO at a reported $380 billion valuation.

Anthropic recently experienced an accidental leak of internal documents about its most powerful model, called Capybara. The disclosure arrived as Bloomberg reported the company is eyeing an October IPO at a reported $380 billion valuation. The leaked draft framed Capybara as a generational advance over Claude Opus 4.6 and described capabilities in coding, academic reasoning, and cybersecurity. The timing and content of the leak have drawn attention to how product cadence and public communication intersect with plans for a potential listing. What we know: - Bloomberg reported Anthropic is targeting an October IPO with a reported $380 billion valuation. - An internal leak described the Capybara model as significantly ahead of Claude Opus 4.6 in coding, reasoning, and cyber capabilities. - The leaked documents said Capybara posed "unprecedented" risks and that it was "currently far ahead of any other AI model in cyber capabilities." - Anthropic has released or previewed multiple Claude iterations rapidly over the past year and is presenting Capybara above the Opus line in a restructured hierarchy. - The company plans an early-access programme giving a small group of organisations advance sight of the model, which observers note may serve both safety testing and enterprise referencing ahead of an IPO. Summary: The leak has made Anthropic's advanced model pipeline publicly visible while also focusing attention on safety disclosures and corporate timing as the company contemplates a public offering. The documents suggest Anthropic is addressing model risks even as it highlights technical capabilities that will invite investor and regulatory scrutiny. Undetermined at this time.