Unverified85% confidenceFactTime unknown

GPT-5.4在OS World Verified测试中达到了75%的得分，首次超越人类的72.4%基准线

1

Sources

85%

Confidence

Medium-term (~90 days)

Relevance

6/1/2026

First Seen

Valid until: 8/30/2026

Sources

GPT-5.4深度评测：编程能力飞跃式提升，但这个缺陷不能忍

bilibilikate人不错

Related Entities

GPT-5.4model OS World Verifiedconcept

Related Claims

Unverified在标准化测试中，GPT-4的表现已超过90%的人类考生78% similar Unverified在DeepSWE基准测试中，GPT 5.5以约70%的通过率高居榜首，比Opus 4.7领先超过15个百分点78% similar 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%77% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/19302

API

curl https://kongchang.com/api/v1/knowledge/claims/19302

MCP

get_claim(id=19302)