Alibaba Launches AI Model That Can Operate PCs and Smartphones

Alibaba’s Qwen division has unveiled the Qwen2.5-VL family of models, capable of managing PCs and smartphones, similar to OpenAI’s Operator virtual assistant.

Following the breakthroughs of China’s DeepSeek, this project marks another significant milestone in AI innovation. According to developers, Qwen2.5-VL outperforms American models, including OpenAI’s GPT-4o, Anthropic’s Claude 3.5 Sonnet, and Google’s Gemini 2.0 Flash, in tasks such as document analysis, solving complex math problems, understanding video content, and answering intricate questions. The model can also extract data from charts and graphs, study multi-hour videos, and recognize scenes from movies and TV shows.

Qwen2.5-VL is accessible through the Alibaba Qwen Chat app or downloadable via Hugging Face. However, developers emphasize that, like many other Chinese AI projects, the model avoids commenting on political issues related to Beijing.