Unverified50% confidenceBenchmarkExact time
Green & Bavelier（2006）研究表明游戏玩家在'有用视野'测试中的表现可提升约30%
1
Sources
50%
Confidence
Long-term
Relevance
8/19/2026
First Seen
Sources
Related Claims
UnverifiedArtificial Analysis给Muse Glimmer 30B打出35分的智能指数，相比Llama 4 Maverick是巨大飞跃60% similarUnverified引导模式下平均端到端准确率从28.6%提升到33.7%，歧义检测F1值从45.3%提升到64.9%56% similarUnverified良好判断团队的预测准确率持续以约30%的优势领先其他专业情报分析机构56% similarUnverified人类队伍的Hit Rate为38.5%（25个Binder），Agent的Hit Rate为34.3%（12个Binder），Fisher精确检验P值为0.83，Mann-Whitney检验P值为0.75，两者无统计显著差异56% similarUnverifiedClaude Sonnet 4.6在智能体工具使用测试中从43.8%提升至61.3%55% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/771888API
curl https://kongchang.com/api/v1/knowledge/claims/771888MCP
get_claim(id=771888)