Qwen-AgentWorld: Language World Models for General Agents

iilreb about 6 hours ago 20 commentsRead Article on arxiv.org

HI version is available. Content is displayed in original English for accuracy.

⚡ Community Insights

Discussion Sentiment

67% Positive

Analyzed from 739 words in the discussion.

Discussion (20 Comments)Read Original on HackerNews

Xx_crazy420_xX•9 minutes ago

I think open-ended simulation for agents will be a key component for training and planning. Similar as human dreams simulate different scenarios in our head. Biggest challenge will be simulating more abstract and complex systems.

Few months ago I did experiment with an open-ended world simulation for AI agent, where the simulated world was progressively building itself based on each of agent actions in open-ended manner. The idea was to give an agent infinite possibility regarding tool calling, where the tool call would be approved by the adjudicator, and the world state would change. The key issues with the PoC were:

  - World decoherence (tried to solve that with a poor graph implementation)
  - World flatness - high abstraction did not account for small events that would compound in real world
  - Start with empty context was real issue to get the agent to explore the world

Anyways the project came to be really funny when you watched agent struggling in desperation to perform real world actions which would be impossible in real world. Main observation was that when presented agent with current action budget, it modulated the creativity and how desperate its actions were.

zkmon•9 minutes ago

What if they did this using GLM 5.2? This looks like a new direction for AI.

adrian_b•38 minutes ago

The smaller of the two models is open weights and available on Huggingface:

https://huggingface.co/Qwen/Qwen-AgentWorld-35B-A3B

blurbleblurble•about 2 hours ago

This might be pretty big. One of my biggest frustrations with smaller models (especially MoE) is their failure to track workflow state at a high level. I'm constantly reminding them what we decided on or asking them to revisit, and reminding them eats context.

Seems like this might make that a lot less painful. And if not off the bat, with some minimal tuning or even just good prompting.

dippogriff•about 2 hours ago

I'm a fan of this direction. For me the most interesting use case for these world models isn't even training, it's verification. If this thing or some idealized version of it can actually reliably simulate state transitions, could you use it to verify an agent's execution path against hard constraints and replace/eclipse LLMs-as-a-judge?

aliljet•44 minutes ago

The benchmarks here are confusing at best. Am I reading correctly that this model is essentially as good or better than all frontier models right now?

anana_•25 minutes ago

I believe the benchmark listed is about simulating the environment for the various tasks, rather than doing them. It seems that the point of this model is to generate sim data to improve other models with

blourvim•37 minutes ago

Benchmarks in general are a little iffy, the whole industry is going off of vibes anyways. Can't decide before trying it out

ElenaDaibunny•43 minutes ago

10M trajectories, probably more of a data scale win than a world model breakthrough tbh

psc007•about 3 hours ago

Eli5? What is this compared to a regular llm assistant model like the base qwen?

gavmor•about 2 hours ago

A regular LLM acts as a "policy," mapping a current state to a specific action (states → actions). Their new LLM acts as a "world model," mapping a current state and a chosen action to a predicted future state ((states, actions) → subsequent states). Instead of deciding "what to do," its explicit objective is to predict the exact environment observation that will result from the interaction history and the agent's current action.

I assumed at first that it was trained on synthetic data, but they actually went and deployed real physical hosts and virtual machines (e.g. Ubuntu, macOS, and Android) and browsers. They ran agentic systems on these continuously and recorded the actual, real-world interactions.

So it's an LLM that infers next state, or outcome,as structured data e.g. literal HTML code, UI view hierarchies, or accessibility trees.

dmos62•21 minutes ago

So, if I'm reading this correctly, whereas a regular LLM would, given a prompt to edit a file, infer a sed call, this "world" model infers the resulting contents of the file.

kakugawa•11 minutes ago

Here's the demo: https://docs.qwenlm.ai/resources/mlu56_demo.html

Here's the description of the world model prompt for the web domain: "A precise GUI state simulator — given the current screen (as HTML) and a user action, predicts the exact next screen as a complete, self-contained HTML document." (You can click the world model prompt box to expand it and see the full prompt.)

So the world model generates the current state (an html document), an agent tells it what action it wants to perform, the world model generates the next state (another html document).

The other domains are similar, but w/ domain-specific nuance.

Tepix•about 3 hours ago

The labels of the very first chart (figure 1, bottom left) are obviously wrong which casts a doubt on the entire paper.

dudisubekti•about 3 hours ago

This label?

> Figure 1: Overview of Qwen-AgentWorld. Top: Qwen-AgentWorld is a unified native language world model across seven domains. Bottom: We explore two complementary strategies for applying world modeling to enhance language agents (mainly using the 35B-A3B model as agent): Decouple and Unify , where the world model serves as the environment simulator and agent foundation model, respectively.

Where is the mistake?

Tepix•about 2 hours ago

The deltas are wrong.

The bars above the label "Infinite Real-World Envs" show growth for example from approx 42 to 55 but the red label says "+7.1". It's wrong for all of them.

dudisubekti•15 minutes ago

Ah I see. Yeah the graphics are probably AI-generated, and AIs do struggle with unit consistency in charts.

(For another example, the charts in the August 2025 GPT-5 presentation)

yorwba•about 1 hour ago

According to Table 6, it's supposed to be 47.9 to 55.

verdverm•about 4 hours ago

35B model from the qwen-3.5 line

https://github.com/QwenLM/Qwen-AgentWorld

https://huggingface.co/Qwen/Qwen-AgentWorld-35B-A3B

khimaros•about 3 hours ago

unsloth, activate!