Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Dong, Guanting, Lu, Junting, Huang, Junjie, Zhong, Wanjun, Liu, Longxiang, Huang, Shijue, Li, Zhenyu, Zhao, Yang, Song, Xiaoshuai, Li, Xiaoxi, Jin, Jiajie, Zhu, Yutao, Wang, Hanbin, Lei, Fangyu, Luo, Qinyu, Chen, Mingyang, Chen, Zehui, Feng, Jiazhan, Wen, Ji-Rong, Dou, Zhicheng
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911608180047872
author Dong, Guanting
Lu, Junting
Huang, Junjie
Zhong, Wanjun
Liu, Longxiang
Huang, Shijue
Li, Zhenyu
Zhao, Yang
Song, Xiaoshuai
Li, Xiaoxi
Jin, Jiajie
Zhu, Yutao
Wang, Hanbin
Lei, Fangyu
Luo, Qinyu
Chen, Mingyang
Chen, Zehui
Feng, Jiazhan
Wen, Ji-Rong
Dou, Zhicheng
author_facet Dong, Guanting
Lu, Junting
Huang, Junjie
Zhong, Wanjun
Liu, Longxiang
Huang, Shijue
Li, Zhenyu
Zhao, Yang
Song, Xiaoshuai
Li, Xiaoxi
Jin, Jiajie
Zhu, Yutao
Wang, Hanbin
Lei, Fangyu
Luo, Qinyu
Chen, Mingyang
Chen, Zehui
Feng, Jiazhan
Wen, Ji-Rong
Dou, Zhicheng
contents Large language models are increasingly expected to serve as general-purpose agents that interact with external, stateful tool environments. The Model Context Protocol (MCP) and broader agent skills offer a unified interface for connecting agents with scalable real-world services, but training robust agents remains limited by the lack of realistic environments and principled mechanisms for life-long learning. In this paper, we present \textbf{Agent-World}, a self-evolving training arena for advancing general agent intelligence through scalable environments. Agent-World has two main components: (1) Agentic Environment-Task Discovery, which autonomously explores topic-aligned databases and executable tool ecosystems from thousands of real-world environment themes and synthesizes verifiable tasks with controllable difficulty; and (2) Continuous Self-Evolving Agent Training, which combines multi-environment reinforcement learning with a self-evolving agent arena that automatically identifies capability gaps through dynamic task synthesis and drives targeted learning, enabling the co-evolution of agent policies and environments. Across 23 challenging agent benchmarks, Agent-World-8B and 14B consistently outperforms strong proprietary models and environment scaling baselines. Further analyses reveal scaling trends in relation to environment diversity and self-evolution rounds, offering insights for building general agent intelligence.
format Preprint
id arxiv_https___arxiv_org_abs_2604_18292
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence
Dong, Guanting
Lu, Junting
Huang, Junjie
Zhong, Wanjun
Liu, Longxiang
Huang, Shijue
Li, Zhenyu
Zhao, Yang
Song, Xiaoshuai
Li, Xiaoxi
Jin, Jiajie
Zhu, Yutao
Wang, Hanbin
Lei, Fangyu
Luo, Qinyu
Chen, Mingyang
Chen, Zehui
Feng, Jiazhan
Wen, Ji-Rong
Dou, Zhicheng
Artificial Intelligence
Computation and Language
Large language models are increasingly expected to serve as general-purpose agents that interact with external, stateful tool environments. The Model Context Protocol (MCP) and broader agent skills offer a unified interface for connecting agents with scalable real-world services, but training robust agents remains limited by the lack of realistic environments and principled mechanisms for life-long learning. In this paper, we present \textbf{Agent-World}, a self-evolving training arena for advancing general agent intelligence through scalable environments. Agent-World has two main components: (1) Agentic Environment-Task Discovery, which autonomously explores topic-aligned databases and executable tool ecosystems from thousands of real-world environment themes and synthesizes verifiable tasks with controllable difficulty; and (2) Continuous Self-Evolving Agent Training, which combines multi-environment reinforcement learning with a self-evolving agent arena that automatically identifies capability gaps through dynamic task synthesis and drives targeted learning, enabling the co-evolution of agent policies and environments. Across 23 challenging agent benchmarks, Agent-World-8B and 14B consistently outperforms strong proprietary models and environment scaling baselines. Further analyses reveal scaling trends in relation to environment diversity and self-evolution rounds, offering insights for building general agent intelligence.
title Agent-World: Scaling Real-World Environment Synthesis for Evolving General Agent Intelligence
topic Artificial Intelligence
Computation and Language
url https://arxiv.org/abs/2604.18292