Unverified50% confidenceBenchmarkExact time
GPT-4 发布时在 MATH 数据集上的正确率为 42.5%
1
Sources
50%
Confidence
Long-term
Relevance
7/24/2026
First Seen
Sources
DeepSeek Math V2深度解析：开源数学AI的能力天花板
bilibiliAI论文许老师教授7/7/2026
Related Claims
VerifiedGPT-5.4 achieved 57.7% accuracy on SWE-Bench Pro, surpassing GPT-5.3-Codex's 56.8%.79% similarUnverifiedGPT-5.5's score on the GPQA doctoral-level scientific reasoning benchmark improved from 78.5% to 85.6%.76% similarPartially VerifiedOn equivalent tasks, GPT-5.5 uses approximately 72% fewer output tokens than Opus.76% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/609178API
curl https://kongchang.com/api/v1/knowledge/claims/609178MCP
get_claim(id=609178)