A data leak on March 26, 2026, exposed Anthropic's unreleased "Claude Mythos" model, described as their most capable yet with significant advances in coding, reasoning, and cybersecurity, which the company has confirmed is now in early access testing. You can read more about that here: 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 market will resolve to "Yes" if Anthropic releases "Claude Mythos" or a model confirmed to be the same model referenced in the leak described above, and that model is made available to the general public by the listed date (ET). Otherwise, this market will resolve to "No." A qualifying model must be named "Claude Mythos" (e.g., Claude Mythos 1, Claude Mythos 5, Claude Mythos X, would count) or be confirmed to be the same model referenced in the leak by Anthropic or by a consensus of credible reporting. Products labeled as Claude Haiku, Sonnet, and Opus 4.7/5.0 or similar will not count for this market's resolution unless they are confirmed to be the same model referenced in the leak. For this market to resolve to "Yes," the qualifying model (as defined above) must be launched and publicly accessible, including via open beta or open rolling waitlist signups. A closed beta or any form of private access will not suffice. The release must be clearly defined and publicly announced by Anthropic as being accessible to the general public. If a qualifying model is made publicly accessible and labeled with the relevant version name within the company’s official website, this will qualify as “publicly announced”. Labeling errors, placeholder text, or version names displayed on the website that do not correspond to a model that is actually accessible to the general public under the rules will not qualify. The primary resolution source for this market will be official information from Anthropic; however, a consensus of credible reporting will also be used.
