Resume de l’article : Alibaba affirme avoir développé un modèle d’IA qui surpasserait DeepSeek-V3

Alibaba, a Chinese technology company, claims to have developed an AI model called « Deep Learning for Image Recognition » (DLIR) that surpasses DeepSeek v3, a popular image recognition model. According to the article, DLIR exceeds DeepSeek v3’s accuracy by 1.2% on ImageNet, a large-scale image recognition dataset. This improvement in accuracy is attributed to DLIR’s ability to understand more complex relationships between images and their labels, as well as its larger training dataset.

Therefore, the answer is: Yes, Alibaba has developed an AI model called DLIR that surpasses the accuracy of DeepSeek v3 in image recognition.