ScholarPulse 日报 2026-07-11

2026-07-11 学术简报：2 篇。ANet Patu-1协议通过自组织协作使异质AI代理网络的集体价值随规模增长并超越同质强模型网络，揭示协作缩放律而非规模缩放。

今日速览

序号	标题	来源	日期	主题	推荐等级
1	ANet Patu-1: The Value of Connection in the Agent Network	arXiv	2026-07-16	AI-Agent	高
2	OmniaBench: Benchmarking General AI Agents Across Diverse Scenarios	arXiv	2026-07-16	AI-Agent	高

重点论文与技术动态

1. ANet Patu-1: The Value of Connection in the Agent Network

来源：arXiv
日期：2026-07-16
作者/机构：Mu Yuan, Jinke Song, Zhaomeng Zhou, Lan Zhang
主题标签：AI-Agent, arXiv
推荐等级：高
分类：cs.NI, cs.AI

一句话结论

ANet Patu-1协议通过自组织协作使异质AI代理网络的集体价值随规模增长并超越同质强模型网络，揭示协作缩放律而非规模缩放。

核心内容

网络价值取决于连接模式：广播星型（Sarnoff）价值∝N，全连接网格（Metcalfe）∝N²，群体形成网络（Reed）∝2^N，类似互联网网络价值模型。
ANet Patu-1是自组织共识协议，网络连续重组联盟，自适应骑乘所有三种连接模式的上包络线，在O(1)并行共识轮次中实现高效共识。
异质代理网络（最廉价模型）集体价值随N增长并超越同质强模型网络，标志协作缩放律而非规模缩放，体现“出现性”与“反射性”结果。

方法与数据

通过形式化指定协议并推导其复杂性分析价值，避免意见评分，符合分布式算法分析范式。
摘要未明确具体数据集或实验细节。

价值判断

值得关注：异质网络自组织协作实现价值超越，揭示协作缩放规律，为AI代理网络设计提供新范式。
可复用点：ANet Patu-1协议可直接复用于AI代理网络的高效自组织共识设计，无需外部设计提示。
局限/待核查：摘要未明确实验局限或验证场景，需进一步实证检验。

（字数：298）

摘要

The Internet taught us that the value of a network depends on \emph{how} its nodes connect: broadcast stars scale as $V\!\propto\!N$ (Sarnoff), fully-connected meshes as $N^2$ (Metcalfe), and group-forming networks as $2^{N}$ (Reed). We ask the analogous question for networks of AI agents. We model the net value of connection as a function of coordination-group size, derive from it the properties an optimal collaboration protocol must have, and introduce ANet Patu-1 -- a self-organizing consensus protocol in which the network continuously re-forms its own coalitions, adaptively riding the upper envelope of all three regimes at $O(1)$ parallel consensus rounds. To measure value without opinion-grading, we score an emergent protocol by formally specifying it and deriving its complexity, the way distributed algorithms are analyzed. Two results follow. (i)~Emergence -- a crowd of the \emph{cheapest} model, when heterogeneous, starts weak but its collective value compounds with $N$ and \emph{overtakes} a crowd of a far \emph{stronger} model that is homogeneous: a crossover that marks a scaling law for collaboration rather than for scale. (ii)~Reflexivity -- a heterogeneous network, given only its own problem and no design hints, converges on ANet Patu-1 itself, reconstructing the high-dimensional law that governs its own connective value.

2. OmniaBench: Benchmarking General AI Agents Across Diverse Scenarios

来源：arXiv
日期：2026-07-16
作者/机构：Chengyu Shen, Yujie Fu, Gangtao Xin, Yanheng Hou, Wenlong Fei, Guojie Zhu
主题标签：AI-Agent, arXiv
推荐等级：高
分类：cs.CL, cs.AI

一句话结论

OmniaBench为通用AI代理在多样化场景中的能力评估提供了全面基准。

一段话。
该基准通过构建覆盖ToC、ToB、ToE的层次化场景分类，合成1431个任务并引入十维能力评估体系，显著挑战当前前沿模型，揭示其在规划、约束维护等关键能力的持续缺陷。

核心内容

覆盖90个一级和354个二级领域，涵盖ToC、ToB、ToE应用场景，形成层次化分类。
通过DAG、DAG-S、Solver和Program四种路径合成单轮与多轮任务。
引入十维能力分类和八种难度因素，支持细粒度评估与分析。

方法与数据

从应用商店、产品文档、行业资源等来源构建层次化场景分类。
数据集含1431个任务，含644个挑战性子集以降低评估成本并避免污染。

价值判断

值得关注：前沿模型如Claude-Sonnet-5和GPT-5.6-Sol的Overall Pass@1仅58.54%和57.14%，凸显基准挑战性。
可复用点：层次化场景分类法和任务合成框架可推广至其他AI代理评估。
局限/待核查：模型在规划、约束维护及自适应校正方面存在持续缺陷。

摘要

Large language models are increasingly evolving from text generators into general agents capable of understanding user requests, invoking external tools, and completing complex tasks through interaction. However, existing agent benchmarks often focus on limited scenarios, tool ecosystems, or interaction formats, making it difficult to systematically characterize model capabilities across heterogeneous application settings. We introduce OmniaBench, a benchmark for evaluating general agents across diverse scenarios with explicit state spaces. We derive application-oriented scenario knowledge from app stores, product documents, industry resources, Web retrieval, and human refinement, forming a hierarchical taxonomy that spans ToC, ToB and ToE with 90 level-1 and 354 level-2 domains. Based on this taxonomy, we construct executable environments and synthesize single-turn and multi-turn tasks through four complementary routes: DAG, DAG-S, Solver, and Program. OmniaBench further introduces a ten-dimensional capability taxonomy and eight compositional atomic difficulty factors to support fine-grained evaluation and analysis. The resulting dataset contains 1,431 tasks, together with a challenging subset of 644 tasks designed to reduce evaluation cost and mitigate potential contamination of the full set after public release. The bench presents substantial challenges to current frontier models, with even Claude-Sonnet-5 and GPT-5.6-Sol achieving Overall Pass@1 scores of only 58.54 and 57.14, respectively. Further analyses reveal clear differences across domains and capabilities, as well as persistent limitations in planning, constraint maintenance, and adaptive correction. OmniaBench provides a broad and diagnostic benchmark for characterizing the capability boundaries of general agents.