Recently, the Qwen team, a subsidiary of Alibaba, announced a major upgrade to its Qwen Deep Research tool. This feature can be activated in the web version of Qwen Chat (a competitor to ChatGPT), offering users a complete closed-loop experience from research, generation, to publication.

The biggest highlight of this update is that users can not only generate comprehensive research reports with citations, but also create interactive web pages and multi-speaker podcasts with one click. The entire process requires only 1-2 clicks, completing data analysis to multi-format content output.

Tongyi Qwen (2)

The new version of Deep Research is supported by Qwen's self-developed model system, including Qwen3-Coder (code generation), Qwen-Image (image rendering), and Qwen3-TTS (text-to-speech synthesis). Unlike previous open-source models, this version is a proprietary hosted service for Qwen, allowing users to use it directly without configuring any infrastructure.

In the demonstration, Qwen showcased its application in U.S. SaaS market research: the model can actively identify data differences, calculate compound growth rates, and generate visual reports with text and images. Subsequently, users can click "Create" to generate web or podcast versions, achieving a three-dimensional presentation of "text, visuals, and audio."

Industry insiders pointed out that the emergence of Qwen Deep Research marks the transition of AI research assistants from "generating text" to "generating multimodal content," showing broad prospects in intelligent research, education, and media creation fields.