Unverified50% confidenceBenchmarkExact time
In the most difficult single-concept quantitative category, most models' accuracy falls below 30%
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
9/15/2026
First Seen
Valid until: 12/14/2026
Sources
Related Entities
Related Claims
UnverifiedEven the best-performing models can only recover 30% of the evaluated evidence and 25% of the limitations descriptions.75% similarVerifiedFor models with more than 30B parameters, the performance loss caused by 4-bit quantization typically results in a 1–3% drop on benchmarks, but the impact is more significant for smaller models.74% similarUnverifiedOn the rare entity slices newly partitioned by structural metrics, the accuracy of current state-of-the-art systems drops by 15.4% to 39.9%.71% similarUnverified轻量版模型单次推理的算力消耗往往只有旗舰版本的十分之一甚至更低71% similarUnverified一项针对GPT-4的研究显示，对MMLU题目进行语义等价改写后，模型准确率在部分子集上下降超过10个百分点70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/921781API
curl https://kongchang.com/api/v1/knowledge/claims/921781MCP
get_claim(id=921781)