---
title: "Qwen3 Technical Report"
canonical_url: "https://www.modelscope.cn/papers/145214"
md_url: "https://www.modelscope.cn/papers/145214.md"
arxiv_id: 2505.09388
published: 2025-05-14
last_updated: 2025-05-14
authors:
  - "An Yang"
  - "Anfeng Li"
  - "Baosong Yang"
  - "Beichen Zhang"
  - "Binyuan Hui"
  - "Bo Zheng"
  - "Bowen Yu"
  - "Chang Gao"
  - "Chengen Huang"
  - "Chenxu Lv"
  - "Chujie Zheng"
  - "Dayiheng Liu"
  - "Fan Zhou"
  - "Fei Huang"
  - "Feng Hu"
  - "Hao Ge"
  - "Haoran Wei"
  - "Huan Lin"
  - "Jialong Tang"
  - "Jian Yang"
  - "Jianhong Tu"
  - "Jianwei Zhang"
  - "Jianxin Yang"
  - "Jiaxi Yang"
  - "Jing Zhou"
  - "Jingren Zhou"
  - "Junyang Lin"
  - "Kai Dang"
  - "Keqin Bao"
  - "Kexin Yang"
  - "Le Yu"
  - "Lianghao Deng"
  - "Mei Li"
  - "Mingfeng Xue"
  - "Mingze Li"
  - "Pei Zhang"
  - "Peng Wang"
  - "Qin Zhu"
  - "Rui Men"
  - "Ruize Gao"
  - "Shixuan Liu"
  - "Shuang Luo"
  - "Tianhao Li"
  - "Tianyi Tang"
  - "Wenbiao Yin"
  - "Xingzhang Ren"
  - "Xinyu Wang"
  - "Xinyu Zhang"
  - "Xuancheng Ren"
  - "Yang Fan"
  - "Yang Su"
  - "Yichang Zhang"
  - "Yinger Zhang"
  - "Yu Wan"
  - "Yuqiong Liu"
  - "Zekun Wang"
  - "Zeyu Cui"
  - "Zhenru Zhang"
  - "Zhipeng Zhou"
  - "Zihan Qiu"
model_name: Qwen3
model_developer: "Qwen Team"
domain:
  - "自然语言处理"
  - "大语言模型"
  - "混合专家模型"
  - "多语言理解"
  - "推理与代码生成"
type:
  - "自然语言处理"
  - "大语言模型"
  - "混合专家模型"
  - "多语言理解"
  - "推理与代码生成"
  - "Computation and Language"
arxiv_url: "https://arxiv.org/abs/2505.09388"
pdf_url: "https://arxiv.org/pdf/2505.09388"
code_link: "https://github.com/QwenLM/Qwen3"
---

# Qwen3 Technical Report

> In this work, we present Qwen3, the latest version of the Qwen model family. Qwen3 comprises a series of large language models (LLMs) designed to advance performance, efficiency, and multilingual capabilities. The Qwen3 series includes models of both dense…

「Qwen3 Technical Report」是 ModelScope 魔搭社区收录的论文，arXiv 2505.09388，作者为 An Yang, Anfeng Li, Baosong Yang et al.，发表于 2025-05-14，属于 自然语言处理、大语言模型、混合专家模型 领域。

- **ArXiv**: 2505.09388
- **Published**: 2025-05-14
- **Authors**: An Yang, Anfeng Li, Baosong Yang, Beichen Zhang, Binyuan Hui, Bo Zheng, Bowen Yu, Chang Gao, Chengen Huang, Chenxu Lv, Chujie Zheng, Dayiheng Liu, Fan Zhou, Fei Huang, Feng Hu, Hao Ge, Haoran Wei, Huan Lin, Jialong Tang, Jian Yang, Jianhong Tu, Jianwei Zhang, Jianxin Yang, Jiaxi Yang, Jing Zhou, Jingren Zhou, Junyang Lin, Kai Dang, Keqin Bao, Kexin Yang, Le Yu, Lianghao Deng, Mei Li, Mingfeng Xue, Mingze Li, Pei Zhang, Peng Wang, Qin Zhu, Rui Men, Ruize Gao, Shixuan Liu, Shuang Luo, Tianhao Li, Tianyi Tang, Wenbiao Yin, Xingzhang Ren, Xinyu Wang, Xinyu Zhang, Xuancheng Ren, Yang Fan, Yang Su, Yichang Zhang, Yinger Zhang, Yu Wan, Yuqiong Liu, Zekun Wang, Zeyu Cui, Zhenru Zhang, Zhipeng Zhou, Zihan Qiu
- **Model**: Qwen3
- **Developer**: Qwen Team
- **Domain**: 自然语言处理, 大语言模型, 混合专家模型, 多语言理解, 推理与代码生成
- **ArXiv URL**: https://arxiv.org/abs/2505.09388
- **PDF**: https://arxiv.org/pdf/2505.09388
- **Code**: https://github.com/QwenLM/Qwen3

Source: https://www.modelscope.cn/papers/145214

---

> Qwen3：通义千问新一代大型语言模型的突破性进展

## 摘要

本文介绍了Qwen3，这是通义千问模型家族的最新版本。Qwen3系列包括一系列大型语言模型（LLMs），旨在提升性能、效率和多语言能力。该系列包含密集型和混合专家（MoE）架构的模型，参数规模从0.6亿到235亿不等。Qwen3的关键创新在于将思考模式（用于复杂多步推理）和非思考模式（用于快速上下文驱动响应）集成到统一框架中，从而无需在不同模型之间切换，并能够根据用户查询或聊天模板动态切换模式。此外，Qwen3引入了思考预算机制，允许用户在推理过程中自适应分配计算资源，以根据任务复杂性平衡延迟和性能。通过利用旗舰模型的知识，显著减少了构建小型模型所需的计算资源，同时确保其竞争力。实验评估表明，Qwen3在代码生成、数学推理、代理任务等多个基准测试中取得了最先进的结果，与更大的MoE模型和专有模型竞争。相比前代Qwen2.5，Qwen3将多语言支持从29种扩展到119种语言和方言，增强了跨语言理解和生成能力。为了促进可重复性和社区驱动的研究开发，所有Qwen3模型均在Apache 2.0许可下公开可用。

## Abstract

In this work, we present Qwen3, the latest version of the Qwen model family. Qwen3 comprises a series of large language models (LLMs) designed to advance performance, efficiency, and multilingual capabilities. The Qwen3 series includes models of both dense and Mixture-of-Expert (MoE) architectures, with parameter scales ranging from 0.6 to 235 billion. A key innovation in Qwen3 is the integration of thinking mode (for complex, multi-step reasoning) and non-thinking mode (for rapid, context-driven responses) into a unified framework. This eliminates the need to switch between different models--such as chat-optimized models (e.g., GPT-4o) and dedicated reasoning models (e.g., QwQ-32B)--and enables dynamic mode switching based on user queries or chat templates. Meanwhile, Qwen3 introduces a thinking budget mechanism, allowing users to allocate computational resources adaptively during inference, thereby balancing latency and performance based on task complexity. Moreover, by leveraging the knowledge from the flagship models, we significantly reduce the computational resources required to build smaller-scale models, while ensuring their highly competitive performance. Empirical evaluations demonstrate that Qwen3 achieves state-of-the-art results across diverse benchmarks, including tasks in code generation, mathematical reasoning, agent tasks, etc., competitive against larger MoE models and proprietary models. Compared to its predecessor Qwen2.5, Qwen3 expands multilingual support from 29 to 119 languages and dialects, enhancing global accessibility through improved cross-lingual understanding and generation capabilities. To facilitate reproducibility and community-driven research and development, all Qwen3 models are publicly accessible under Apache 2.0.