Matthew Berman tests an interactive Rubik's Cube, a Fall Guys-style game, an ocean scene and a watermelon simulation. The cube behaves convincingly in his demonstration, while the other examples show flickering, clipping, shadow problems and limited realism rather than uniformly polished results.
Matthew Berman reports repeated quota problems while moving between Muse Code plans. He also highlights an unrelated, puzzling model response during the boat task without establishing its cause or broader significance. These are early creator observations, not a comprehensive or independently reproduced evaluation.
Matthew Berman separates the API-accessible Muse Wonder from subscription-only Wonder Full when reviewing reported benchmark and cost charts. Strong coding and aggregate scores are useful context, but he still weighs practical output, access limits and cost per completed task rather than declaring a universal winner.
Watch on YouTube
