Anthropic Just Leaked Upcoming Model With “Unprecedented Cybersecurity Risks” in the Most Ironic Way Possible

Anthropic has confirmed it's developing a significantly more powerful model called Claude Mythos.

The company says Claude Mythos represents a major step forward and is the most capable model they've built so far.

A leaked draft blog warned the model could create cybersecurity risks not seen before and said results would be shared to help defenders prepare.

Leaks place Claude Mythos in a new 'Capybara' tier, which fits alongside Opus, Sonnet, and Haiku in decreasing size, capability, and cost.

On benchmarks, Capybara models score higher than Claude Opus 4.6 in software coding, academic reasoning, and cybersecurity tests.