Claude Opus 5.5 and Opus 5 solve two tasks, miss one

Both models solved a logic grid and stone game and missed an ordering puzzle in an API comparison; Opus 5.5’s output rate was about 11% higher.