Unverified50% confidenceBenchmarkExact time
移除歧义后所有模型准确率提升26.8至40.2个百分点，Gemini甚至达到81%，表明歧义是深度搜索代理最大性能瓶颈之一
1
Sources
50%
Confidence
Long-term
Relevance
7/9/2026
First Seen
Sources
Related Claims
Unverified可变粒度搜索（Variable Granularity Search）相比束搜索和DVTS等SOTA方案，实现约3%的准确率提升，同时减少52%的计算量70% similarUnverified研究数据表明，在数学竞赛级别问题上，启用深度推理的模型准确率可从不足30%提升至80%以上68% similarUnverifiedIn vertical domains such as industrial defect detection and pathology slide analysis, the zero-shot detection accuracy of large models such as GPT-4V and Gemini lags behind specialized models by 20–40 percentage points on the mAP metric.68% similarUnverifiedAccording to the LongMEM Eval benchmark, AgentMemory's retrieval accuracy reaches 95.2%, surpassing the similar project Mem0 by 27 percentage points.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/387245API
curl https://kongchang.com/api/v1/knowledge/claims/387245MCP
get_claim(id=387245)