Unverified50% confidenceFactExact time
基准污染（Benchmark Contamination）导致HumanEval、MMLU等主流基准区分度下降，部分模型HumanEval得分超过95%、MMLU超过90%但实际能力平庸
1
Sources
50%
Confidence
Long-term
Relevance
7/24/2026
First Seen
Sources
Qwen3 235B A22B实测：2.4万亿参数开源模型超越Claude Opus
Related Claims
UnverifiedSome models have already surpassed 90% of the human average score on MMLU, causing their discriminative power to drop sharply.80% similarUnverifiedBy mid-2024, top models had generally exceeded scores of 85–90% on MMLU, causing the benchmark's discriminative power to drop sharply.73% similarUnverifiedTraditional benchmarks such as MMLU and HumanEval have experienced benchmark saturation due to training data contamination; when GPT-4 was released, its MMLU score was approximately 86%, whereas multiple open-source models have now surpassed 90%73% similarUnverifiedSince the second half of 2024, the score improvements of mainstream large models on standard benchmarks such as MMLU and HumanEval have narrowed noticeably.72% similarUnverifiedOn mainstream evaluations such as MMLU, HumanEval, and GSM8K, the score gaps among leading large models are often only a few percentage points.69% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/602713API
curl https://kongchang.com/api/v1/knowledge/claims/602713MCP
get_claim(id=602713)