Anthropic accidentally exposed details about Claude Mythos, described as its most powerful AI model to date, through an unsecured data cache on March 26. The company confirmed the model represents a "step change" in capabilities across coding, reasoning, and cybersecurity.
For the broader landscape, see our complete guide to AI coding tools in 2026.
What Happened
Security researchers Roy Paz and Alexandre Pauwels discovered roughly 3,000 unpublished assets in a publicly searchable data store belonging to Anthropic. Among the exposed files was a draft blog post detailing Claude Mythos, internally codenamed "Capybara," a new model tier positioned above the current Claude Opus 4.6.
The leak resulted from Anthropic's content management system defaulting assets to public unless explicitly set to private. Fortune first reported the exposure after the researchers flagged it. Anthropic attributed the incident to "human error" in CMS configuration.
Leaked documents show Mythos outperforms Opus in software coding, academic reasoning, and cybersecurity tasks. An Anthropic spokesperson confirmed the model is "currently far ahead of any other AI model in cyber capabilities." The company is rolling it out to early access customers with a focus on helping cybersecurity defenders first.
Why It Matters for Creators
If you use Claude for writing, coding, or creative workflows, Mythos signals a meaningful jump in what AI assistants can handle. Better reasoning means fewer hallucinations and more reliable outputs when building complex creative projects. The coding improvements suggest tools built on Claude, including Marketplace integrations, will become substantially more capable once Mythos reaches general availability.
There is no public release date yet. Leaked documents indicate Mythos will be "very expensive for us to serve, and will be very expensive for our customers to use," with Anthropic working on efficiency improvements before general release. The company is taking a phased approach, starting with enterprise customers before broader access. Creators relying on Claude should expect Mythos to arrive after its safety review and controlled rollout are complete.
What to Do Next
No public access is available yet. Watch Anthropic's news page for official announcements. Enterprise Claude customers can contact their account representatives about early access. For now, Claude Opus 4.6 remains the top-tier option while Mythos works through its phased release.
This story was covered by Creative AI News.
Subscribe for free to get the weekly digest every Tuesday.
