研究人员推出了一项名为DeepWeb-Bench的新基准，旨在评估前沿语言模型的深度研究能力。该基准比现有基准更具挑战性，需要广泛收集证据、跨源核对和多步推理。对九个前沿模型的初步评估显示，推理和校准失败是主要瓶颈，占错误总数的70%以上，而非检索问题。 AI
影响 该基准将推动前沿模型在复杂推理和证据综合方面取得进步，超越简单的检索任务。
排序理由 该集群描述了一篇在arXiv上发布的新学术基准论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2605.21482v1 Announce Type: new Abstract: Deep research, in which an agent searches the open web, collects evidence, and derives an answer through extended reasoning, is a prominent use case for frontier language models. Frontier deep research products score high on existin…
Deep research, in which an agent searches the open web, collects evidence, and derives an answer through extended reasoning, is a prominent use case for frontier language models. Frontier deep research products score high on existing benchmarks, making it difficult to distinguish…