Structured pruning usually causes accuracy degradation in large models, but a drop in pass@1 does not necessarily mean reasoning ability is impaired; it may simply be that correct answers are harder to sample.