Alibaba has introduced a cutting-edge artificial intelligence model capable of comprehending images and engaging in more intricate conversations than their prior offerings, intensifying the competitive global race for technological supremacy.
Alibaba, the Chinese tech giant, is introducing two new open-source models, Qwen-VL and Qwen-VL-Chat, allowing global researchers, academics, and companies to develop AI applications without the need for extensive training, saving time and costs. Qwen-VL specializes in answering open-ended queries about various images and generating image captions.
On the other hand, Qwen-VL-Chat is designed for more intricate interactions, as Alibaba explains. It can handle tasks like analyzing multiple image inputs, responding to multiple rounds of questions, crafting stories, generating images from user-provided photos, and solving mathematical equations depicted in pictures.
Alibaba showcased an example where their AI can interpret a Chinese hospital sign and answer questions about department locations. While generative AI has mostly concentrated on text-based responses, advancements like OpenAI’s ChatGPT now enable understanding images and providing text-based responses, similar to Qwen-VL-Chat.
Alibaba’s new models, Qwen-VL and Qwen-VL-Chat, are extensions of their Tongyi Qianwen language model. Tongyi Qianwen is a large language model used for chatbot applications, and it was released earlier this year. Alibaba is open-sourcing these models, not generating licensing fees, but aiming to expand their user base for these AI models. This strategy aligns with Alibaba’s efforts to boost growth in its cloud division, which is gearing up for a public listing.