Unverified60% confidenceBenchmarkExact time
MMLU scores jumped from approximately 70% for GPT-3.5 to 86% for GPT-4, and then to above 90% for subsequent models.
2
Sources
60%
Confidence
Long-term
Relevance
8/19/2026
First Seen
Sources
Related Claims
UnverifiedGPT-4在2023年初于MMLU上取得86.4%的成绩80% similarUnverifiedModels such as GPT-4, Claude, and Gemini successively broke the 90% accuracy threshold on MMLU, causing the benchmark to lose its discriminative power.77% similarUnverifiedOn mainstream benchmarks such as MMLU, HumanEval, and GSM8K, open-source models have reached 85–95% of the level of early versions of GPT-4.77% similarUnverifiedGPT-4在LeetCode竞赛题上的通过率超过80%73% similarUnverifiedGPT-6 Astra scored a high 90% (72/80) on the KingBench 3 benchmark.73% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/774422API
curl https://kongchang.com/api/v1/knowledge/claims/774422MCP
get_claim(id=774422)