1
ReasonBENCH: Benchmarking the (In)Stability of LLM Reasoning
Benchmark揭示大语言模型推理时的一致性波动，系统性评估稳定性短板，为构建可靠LLM应用提供关键参考。
arXiv:2512.07795v2 Announce Type: replace-cross Abstract: Benchmark scores for LLM reasoning systems are reported as single numbers, yet the same mode…