Last released Sep 23, 2026
Watch the chain of thought of open-weight reasoning models and flag what you worry about.