Unverified50% confidenceFactExact time

顶级自回归模型如GPT-4o、Claude 3.5在GSM8K上准确率已超过95%，导致基准饱和

1

Sources

50%

Confidence

Medium-term (~90 days)

Relevance

7/17/2026

First Seen

Valid until: 10/15/2026

Sources

扩散语言模型推理陷阱：一行代码修复「灵活性」缺陷

Related Claims

VerifiedGPT-5.4在SWE-Bench Pro测试中拿下57.7%准确率，超越了GPT-5.3-Codex的56.8%79% similar UnverifiedGPT-5.5在GPQA博士级科学推理基准上从78.5%提升到85.6%77% similar Unverified经过优化的提示词可以将GPT-4在特定任务上的准确率从60%提升至95%以上77% similar UnverifiedGPT-5.6引入了'UltraFast'模式，推理速度提升2到3倍76% similar Unverified该Skill结合GPT-5.x级别模型，目前稳定性可达95%76% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/543860

API

curl https://kongchang.com/api/v1/knowledge/claims/543860

MCP

get_claim(id=543860)