Alibaba unveils new Tongyi Qianwen AI language model

Alibaba Cloud, the cloud division of Alibaba Group, launched its 72-billion-parameter version of Tongyi Qianwen, an AI language model code-named Qwen-72B, on December 1, 2023. The model demonstrates superior performance over OpenAI’s GPT-3.5 and GPT-4 in English and Chinese language understanding, mathematical reasoning, and coding tasks. Alibaba Cloud also released Qwen-1.8B, a smaller model, and Qwen-Audio, an audio-focused AI model, alongside previously open-sourced versions with 7 billion and 14 billion parameters. Zhou Jingren, Alibaba Cloud’s CTO, highlighted the importance of open-source ecosystems in advancing AI technology in China. He stated the company will continue investing in open-source large language models (LLMs) to foster innovation and accessibility. The goal is to provide the most open LLM platform in the AI era, reducing barriers for enterprises to train and deploy customized AI solutions. The open-source approach aims to simplify model development, lower entry costs, and enable faster adoption of AI applications across industries. Alibaba Cloud’s strategy positions Tongyi Qianwen as a competitive alternative to proprietary models like ChatGPT, supporting China’s growing demand for locally developed AI technologies. The release follows Alibaba’s earlier open-sourcing of smaller LLMs, reinforcing its commitment to collaborative AI development. The company’s focus on accessibility and customization could accelerate AI integration in Chinese businesses and research institutions.