No official torrent yet β requesting it tells the flock what to host next. You cannot seed a catalog-only page.
Cataloged from Qwen/Qwen-AgentWorld-35B-A3B
All models by Qwen β
π Technical Report | π Blog | π€ Hugging Face | π€ ModelScope | π» GitHub | π₯οΈ Demo
[!Note] This repository contains the model weights and configuration files for Qwen-AgentWorld-35B-A3B, a native language world model trained for agentic environment simulation.
These artifacts are compatible with Hugging Face Transformers, vLLM, SGLang, etc.
Qwen-AgentWorld is the first language world model to cover seven agent interaction domains within a single model. It simulates agentic environments via long chain-of-thought reasoning, predicting the next environment state given an agent's action and interaction history. Trained through a three-stage pipeline β CPT injects environment knowledge, SFT activates next-state-prediction reasoning, RL sharpens simulation fidelity β Qwen-AgentWorld is a native world model: environment modeling is the training objective from the CPT stage onward, not a post-hoc add-on.
Five-dimensional rubric mean per domain, normalized to 0-100 scale.
Qwen-AgentWorld-35B-A3B can be served via APIs with popular inference frameworks. In the following, we show example commands to launch OpenAI-compatible API servers.
[!Important] The model has a default context length of 262,144 tokens. If you encounter out-of-memory (OOM) errors, consider reducing the context window. However, because Qwen-AgentWorld leverages extended context for multi-turn environment simulation, we advise maintaining a context length of at least 128K tokens.
SGLang is a fast serving framework for large language models.
An OpenAI-compatible API will be available at http://localhost:8000/v1.
vLLM is a high-throughput and memory-efficient inference engine for LLMs.
The
--language-model-onlyflag is required because the model architecture includes visual component definitions but the checkpoint only contains language model weights. Without this flag, vLLM will attempt to initialize visual modules and fail.
An OpenAI-compatible API will be available at http://localhost:8000/v1.
[!Note] We provide domain-specific world model system prompt templates in
prompts/of the GitHub repository for all 7 domains. These serve as general-purpose system prompts when using Qwen-AgentWorld as an environment simulator. Each domain folder contains asystem_prompt.txt(world model system prompt) and ajudge_system_prompt.txt(evaluation prompt).
AgentWorldBench evaluates language world models by scoring each predicted environment observation on 5 dimensions: Format, Factuality, Consistency, Realism, and Quality.
The evaluation follows a three-step pipeline:
Sampling Parameters: We recommend temperature=0.6, top_p=0.95, top_k=20 for world model inference. The model uses thinking mode by default (<think>...</think>) to reason about environment state transitions before producing the predicted observation.
Adequate Output Length: We recommend an output length of 32,768 tokens for most queries. For long, multi-step trajectories, you may increase the max output length to accommodate detailed environment observations.
Domain-Specific System Prompts: For optimal simulation fidelity, use the domain-specific system prompts provided in the prompts/ directory of the GitHub repository.
If you find our work helpful, feel free to give us a cite.
python -m sglang.launch_server \ --model-path Qwen/Qwen-AgentWorld-35B-A3B \ --port 8000 \ --tp-size 4 \ --context-length 262144 \ --reasoning-parser qwen3vllm serve Qwen/Qwen-AgentWorld-35B-A3B \ --port 8000 \ --tensor-parallel-size 4 \ --max-model-len 262144 \ --reasoning-parser qwen3 \ --language-model-only \ --trust-remote-codefrom transformers import AutoModelForCausalLM, AutoTokenizer
model_name = "Qwen/Qwen-AgentWorld-35B-A3B"tokenizer = AutoTokenizer.from_pretrained(model_name)model = AutoModelForCausalLM.from_pretrained( model_name, torch_dtype="auto", device_map="auto",)
messages = [ { "role": "system", "content": "You are a language world model simulating a Linux terminal environment. " "Given the user's command, predict the terminal output." }, { "role": "user", "content": "Action: execute_bash\nCommand: ls -la /home/user/project/" }]
text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)inputs = tokenizer([text], return_tensors="pt").to(model.device)outputs = model.generate(**inputs, max_new_tokens=2048, temperature=0.6)response = tokenizer.decode(outputs[0][inputs.input_ids.shape[-1]:], skip_special_tokens=True)print(response)from openai import OpenAI
client = OpenAI( base_url="http://localhost:8000/v1", api_key="EMPTY",)
# Terminal domain examplemessages = [ { "role": "system", "content": "You are a language world model simulating a Linux terminal environment. " "Given the user's command, predict the terminal output." }, { "role": "user", "content": "Action: execute_bash\nCommand: ls -la /home/user/project/" }]
response = client.chat.completions.create( model="Qwen/Qwen-AgentWorld-35B-A3B", messages=messages, max_tokens=32768, temperature=0.6,)print(response.choices[0].message.content)# Clone the evaluation repositorygit clone https://github.com/QwenLM/Qwen-AgentWorld.gitcd Qwen-AgentWorld
# Download the benchmarkhuggingface-cli download Qwen/AgentWorldBench --repo-type dataset --local-dir ./AgentWorldBench
# Install dependenciespip install openaicd eval
# Step 1: Run world model inferencepython eval.py infer \ --data-dir ../AgentWorldBench \ --model-base-url http://localhost:8000/v1 \ --model-name Qwen/Qwen-AgentWorld-35B-A3B \ --output-dir ./results
# Step 2: Run LLM judge scoringexport OPENAI_API_KEY="your-api-key"python eval.py judge \ --predictions ./results/predictions.jsonl \ --judge-base-url https://api.openai.com/v1 \ --judge-model gpt-5.2-2025-12-11 \ --output-dir ./results
# Step 3: Aggregate and display scorespython eval.py score --predictions ./results/judged.jsonl@article{zuo2026qwen, title={Qwen-agentworld: language world models for general agents}, author={Zuo, Yuxin and Xiao, Zikai and Sheng, Li and Huang, Fei and Tu, Jianhong and Liu, Yuxuan and Tang, Tianyi and Hu, Xiaomeng and Su, Yang and Lan, Qingfeng and others}, journal={arXiv preprint arXiv:2606.24597}, year={2026}}