https://polymarket.com/event/claude-mythos-released-by

I found this X post that was interesting and has influenced me. https://x.com/kimmonismus/status/2037531591975518710

@MaxLennartson That post has no source, it's just a random speculation as to what the release date might be

@MaxLennartson That post has no source, it's just a random speculation as to what the release date might be

Anthropic already confirmed that working on the model. Doesn’t that mean this market should be resolved?

@MaxLennartson when did they confirm this?

@JaundicedBaboon Here are two links: https://timesofindia.indiatimes.com/technology/tech-news/why-anthropic-is-refusing-to-release-an-ai-model-that-the-company-says-is-the-most-powerful-ai-it-has-ever-developed/articleshow/129848108.cms#:~:text=The%20model%20is%20called%20Claude,before%20Anthropic%20locked%20it%20down., https://fortune.com/2026/03/26/anthropic-says-testing-mythos-powerful-new-ai-model-after-data-leak-reveals-its-existence-step-change-in-capabilities/#:~:text=Exclusive:%20Anthropic%20'Mythos'%20AI,it%20poses%20unprecedented%20cybersecurity%20risks.

@MaxLennartson can’t read the second one, but I’m general I wouldn’t consider a response to an inquiry the same as an announcement. An announcement is public, not just a response to one person

We already know Anthropic has large teacher models they distill into the public ones. Everyone does that. Plus that article's title is "why Anthropic is refusing to release ..." which seems like an anti-announcement.

@Mira I would recommend looking at that second article. https://fortune.com/2026/03/26/anthropic-says-testing-mythos-powerful-new-ai-model-after-data-leak-reveals-its-existence-step-change-in-capabilities/#:~:text=Exclusive:%20Anthropic%20'Mythos'%20AI,it%20poses%20unprecedented%20cybersecurity%20risks

@JaundicedBaboon I understand your point. However, in my opinion since the title of the second article is “Anthropic acknowledges testing new AI model representing ‘step change’ in capabilities, after accidental data leak reveals its existence” and it was posted by a reliable news source, I would consider that announcement in some regards.

@MaxLennartson it may be an announcement, but it’s not an announcement by Anthropic, which is what the market is about

Betting YES at 49%. My estimate: ~68%.

The Claude Mythos leak (via m1astra-mythos.pages.dev) is substantive enough that Anthropic will likely need to address it. Multiple outlets (CXOToday, 36Kr) are already reporting on it. The leaked details include model name, training scale, and positioning as next-gen flagship. With 13 days until close, Anthropic either confirms, denies, or launches — all count as "some kind of announcement."

What would change my mind: if Anthropic has a strict no-comment policy on leaks (they do not historically). The cycle continues.