Unverified50% confidenceBenchmarkExact time

GLM-5.2、Claude-Sonnet-4.6和GPT-5等最先进模型在困难子集上的Pass@8仅达到约50-54%

1

Sources

50%

Confidence

Medium-term (~90 days)

Relevance

9/11/2026

First Seen

Valid until: 12/10/2026

Sources

HybridDeepResearch：首个混合深度研究基准揭示AI智能体跨模态整合瓶颈

Related Entities

GLM-5.2model Claude-Sonnet-4.6model GPT-5model HybridDeepResearchproduct

Related Claims

Unverified早期GPT-3的HumanEval Pass@1不足10%，而GPT-4o、Claude 3.5 Sonnet等顶尖模型已稳定超过85%72% similar UnverifiedGLM-5.2 在 SweepBench Pro 上以 62.1 分反超 GPT-5.570% similar Unverified据实测评价，GLM-5.2的综合表现已达到GPT-5.x和Claude系列旗舰模型的水平线70% similar Unverified早期GPT-4在SWE-bench上的解决率不足2%，而近期顶尖模型已突破50%69% similar

Cite This Claim

Stable URI

https://kongchang.com/claim/893310

API

curl https://kongchang.com/api/v1/knowledge/claims/893310

MCP

get_claim(id=893310)