Unverified50% confidenceBenchmarkExact time
Meituan LongCat launched the Lohosearch search agent benchmark, where the best score among 11 frontier models tested was only 34.74%, far below the approximately 90% human performance.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
7/22/2026
First Seen
Valid until: 10/20/2026
Sources
DeepSeek V4灰度测试开启，Grok 5.0蓄势待发｜AI大模型最新动态
Related Claims
UnverifiedMultiple benchmark tests in 2023 (such as the HaHackathon dataset) showed that top LLMs still have significantly lower accuracy than ordinary human annotators on sarcasm recognition tasks70% similarUnverifiedMainstream benchmarks for evaluating Agent planning capabilities include ALFWorld, WebArena, and GAIA, and the scores of current state-of-the-art models remain far below human-level performance.68% similarUnverified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一66% similarUnverifiedAccording to the LongMEM Eval benchmark, AgentMemory's retrieval accuracy reaches 95.2%, surpassing the similar project Mem0 by 27 percentage points.65% similarUnverified谷歌DeepMind团队在2022年的研究中记录了超过150种在小模型中缺失、却在大模型中出现的能力，包括多步推理、类比迁移与少样本学习63% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/592565API
curl https://kongchang.com/api/v1/knowledge/claims/592565MCP
get_claim(id=592565)