ARC-AGI-3 tests whether models can reason through novel problems, not just recall patterns, a task even top systems still ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible resultsSome results have been hidden because they may be inaccessible to you
Show inaccessible results