Model 2 is the name under which Anthropic’s August 2026 Risk Report discloses a model it does not intend to release, more powerful than the shipped Claude Mythos 5 and carrying a misalignment-risk rating nudged from very low to low. In this corpus it exists only through that report and the arithmetic done on it: it beat Mythos 5 by 12.5 points on CoBench v2 in the item where the lab says its AI R&D evaluations have saturated, the clearest instance of the public frontier detaching from the real one.
AISystem
Model 2
Anthropic's unreleased internal model, disclosed in August 2026 as more powerful than Mythos 5 and with no plans for release.
Sources
Attribution and reporting linked by this record; links do not imply independent verification.
- Welcome to August 15, 2026 (theinnermostloop.substack.com)
Development timeline
Newest reporting dates first. Reporting dates may differ from event dates.
Cite or correct this record
Suggest a correction — opens a GitHub draft for you to review and submit.