The model has always reasoned with full thoughts (and continues to do so), now we just have an option to return the summarized thoughts

→ View original post on X — @officiallogank