Will you look at that, Anthropic is planning to release a new class of model that will represent a step-change in coding and reasoning capability.

https://archive.ph/2026.03.27-023147/https://fortune.com/2026/03/26/anthropic-says-testing-mythos-powerful-new-ai-model-after-data-leak-reveals-its-existence-step-change-in-capabilities/

This is obviously unverifiable and very easy to fake, but supposedly someone saved the leaked post before it was taken down.

Good bye 👋🏻

https://m1astra-mythos.pages.dev/

Claude Mythos

@kyle I love the complete absence of usable information here. Very big! Very expensive! Very scary!
@stevestreza when I think “capybara” I think of the deep connective tissue that links together knowledge and ideas
@stevestreza @kyle when they say they’re scared of their own creation that’s how you know the model’s really good
@harpaa01 @stevestreza @kyle Claude Ballarina Cappuccina is our new model and is far ahead of any other models in the brain rot class