Model Releases

Qwen-AgentWorld: a language world model for general-purpose agents

Original title:Qwen-AgentWorld: Language World Models for General Agents

AI summary

The Qwen team released Qwen-AgentWorld, a native language world model trained with environment modeling as its objective, simulating seven domains in a single model: MCP, Search, Terminal, SWE, and GUI (Web, OS, Android). The model was trained on over 10 million real interaction trajectories and achieves the highest simulation quality on AgentWorldBench with the Qwen-AgentWorld-397B-A17B version, surpassing GPT-5.4, Claude Opus 4.8, and Gemini 3.1 Pro. The team also released the AgentWorldBench evaluation benchmark. The model can serve as a decoupled environment simulator for agent RL training, or as a unified agent foundation model that transfers without agent RL fine-tuning after LWM warm-up. The model and benchmark are open-source on Hugging Face and ModelScope.

Key Takeaways

  • The Qwen team released Qwen-AgentWorld,a native language world model trained with environment…
  • The model was trained on over 10 million real interaction trajectories and achieves the highest…
  • 4,Claude Opus 4.
  • 8,and Gemini 3.
  • The team also released the AgentWorldBench evaluation benchmark.

Quick Facts

Published
Updated
Source
Qwen:Blog Retrieval(API)
Category
Model Releases
AI score
81
Read the original Read the full article at the original source
Why it matters

Qwen把世界模型做成了一个可开源的通用产品,覆盖七域,做agent RL的可以直接拿它仿真训练,可控性甚至超过真实环境,做agent的团队应该认真看看。

Related entities

Source & copyright

This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.

Original article · AIHOT entry

How to Cite

When citing this article, keep the title, date, and permalink, and credit Orange News as the source.

Orange News: Qwen-AgentWorld: a language world model for general-purpose agents. June 23, 2026. https://new.nat6.net/en/news/cmqrjhuo40kszslp5qsjmxdy7/

View all →