China-based Alibaba rolls out two open-source AI models, the Qwen-7B and Qwen-7B-Chat. These models are provided for developers of individuals and small and medium-sized enterprises free of license fees.
It marks the first time a Chinese tech giant has released an open-sourced Large Language Model(LLM), which is trained on vast amounts of data and serves as the basis for generative AI apps like ChatGPT. The release comes after Meta’s Llama2 open-source AI model was unveiled last month.
With this, Alibaba and Meta’s AI models can chip away at the current market dominance of OpenAI’s viral AI chatbot, the ChatGPT, which charge users expensive fees to generate human-like responses to user prompts.
Pundits believe the distribution will help Alibaba secure a wide base of users including researchers, the academic field, and businesses around the world to develop their own AI apps. It could save their time and expense from needing to train their own systems. Royalties could be charged to big corporations with more than 100 million monthly active users.
This comes at a time when Alibaba is looking to boost its cloud computing division through investments in AI for profitability and future growth. In April, Alibaba Cloud, its cloud computing arm, unveiled its ChatGPT-style product, the Tongyi Qianwen, which allows AI content generation in English and Chinese.
Similar to the Qwen-7B, Tongyi Qianwen has seven billion parameters, the basic metric used to measure the performance of each AI Model. This is one-tenth that of Meta’s first open-source AI model, the Llama2 unveiled last month. The Qwen-7B-Chat, a version designed for conversational apps, will have more than seven billion parameters.