Alibaba has introduced an updated version of its artificial intelligence (AI) model, Tongyi Qianwen 2.0, to vie with tech giants like Amazon and Microsoft. This iteration represents the latest large language model (LLM) from China’s leading cloud computing and e-commerce enterprise and is the basis for AI-powered chatbots.
Alibaba emphasized that the new model, Tongyi Qianwen 2.0, is a significant enhancement from its earlier version launched in April of the same year. With “hundreds of billions of” parameters, a metric to gauge AI model power, the new model boasts one of the world’s most potent AI models by parameter count.
Compared to other models, Tongyi Qianwen 2.0 surpasses Tencent’s Hunyuan AI, Microsoft-backed OpenAI’s GPT-3, and Meta Platform’s Llama 2 in terms of parameter counts. Alibaba affirmed that Tongyi Qianwen 2.0 exhibits advanced abilities in comprehending intricate instructions, copywriting, logical reasoning, memorization, and preventing misinformation—referred to as hallucinations in the context of AI chatbots.
Alibaba emphasized the model’s intended applications in specialized industries such as legal advice and finance. Additionally, the company launched eight AI models designed for diverse sectors, including entertainment and healthcare.
The prior Tongyi Qianwen model was integrated into two solutions—Qwen-VL and Qwen-VL-Chat. Qwen-VL is capable of responding to diverse image-related open-ended queries and generating image captions, while Qwen-VL-Chat can engage in multifaceted interactions, such as analyzing multiple image inputs and answering successive rounds of questions.
In addition to unveiling Tongyi Qianwen 2.0, Alibaba introduced the GenAI Service Platform, enabling companies to develop their generative AI applications using proprietary data. This offering is set to compete with Microsoft’s Azure OpenAI Studio and Amazon Web Service’s Bedrock.
