Chinese e-commerce giant Alibaba’s Qwen team of researchers and AI model engineers has rolled out a significant extension to its Qwen Deep Research tool, which is available as an optional method that a user can activate in Qwen’s online chat (a ChatGPT competitor).
The update enables users to generate not only comprehensive research reports with well-organized citations, but also interactive websites and podcasts with multiple speakers – all in 1-2 clicks.
This functionality is part a writer’s editiondifferent from many of Qwen’s previous open source model offerings.
Although the feature is based on open source models Qwen3 encoder, Image by QwenAND Qwen3-TTS To meet your core capabilities, you need end-to-end experience – spanning research execution, network deployment, and audio generation run and operated by Qwen.
This means users can profit from a managed, integrated workflow without having to configure infrastructure. That said, developers with access to open source models could theoretically replicate similar functionality on private or industrial systems.
The update was announced via a team official X account (@Alibaba_Qwen) today, October 21, 2025, stating:
“Qwen Deep Research has just undergone a major update. It now produces not only a report, but also a live website and podcast – powered by Qwen3-Coder, Qwen-Image and Qwen3-TTS. Your insights are now visual and audio.”
Research results in multiple formats
The basic workflow starts with a user request in the Qwen Chat interface. From there, Qwen collaborates by asking clarifying inquiries to shape the scope of the research, pulls data from the Internet and official sources, and analyzes or resolves any inconsistencies found, even generating custom code if needed.
AND demo video posted by Qwen on X walks through this process on Qwen Chat using the example of the American SaaS market.
In it, Qwen mines data from multiple industry sources, identifies discrepancies in market size estimates (e.g., $206 billion vs. $253 billion), and highlights uncertainties about the U.S. share of the global numbers. The assistant comments on differences in scope between sources and calculates a compound annual growth rate (CAGR) of 19.8% from 2020 to 2023, providing contextual evaluation to support the raw numbers.
Once the study is complete, users can click the “eyeball” icon below the output results (see screenshot), which is able to display the report in PDF format on the right panel.
Then, when viewing the report in the right pane, the user can click the “Create” button in the upper right corner and select one of two options:
-
“Website creator” which produces A live, skilled websitemechanically deployed and hosted by Qwenusing Qwen3-Coder for structure and Qwen-Image for visualization.
-
“Podcast” which he states produces sound podcastcontaining dynamic, multi-speaker narration generated by Qwen3-TTS hosted by Qwen for easy sharing and playback.
This enables users to quickly transform a single research project into multiple forms of content – written, visual and audio – with minimal additional input.
The site contains graphics generated by Qwen Image, making it suitable for use in public presentations, classrooms, or publications.
The podcast feature allows users to select from 17 different speaker names as host and 7 as co-host, although I haven’t been able to find a way to preview the voice messages before selecting them. It seems designed for deep listening on the go.
There was no way to change the displayed language, so mine, as well as my reports and initial prompts, were in English, although Qwen LLMs are multimodal. The voices were a bit more automatic than other AI tools I’ve used.
Here is an example of the website I generated on the similarities in authoritarian regimes throughout history, another one about UFO or UAP sightingsand below this paragraph a podcast about UFO or UAP sightings.
While the site is hosted via a public link, the podcast must be downloaded by the user and cannot be publicly linked to, from what I have been able to determine so far in my short use.
Note that the podcast is very different from the actual report – it’s not just a simple audio version of it for you to read, but rather a new format in which two hosts discuss and banter about a topic, using the report as a starting point.
The website versions of the report also include new graphics that are not included in the PDF report.
Comparisons with Google’s NotebookLM
While the new capabilities were well received by many early adopters, comparisons with other research assistants – particularly Google – emerged NotebookLMwhich recently came out of beta.
AI commentator and newsletter writer Chubby (@kimmonismus) noted on X: :
“I really appreciate that Qwen provides regular updates. It’s great.
However, trying to build a NotebookLM clone in Qwen-3-max doesn’t sound very promising compared to Google’s version.”
While NotebookLM is based on organizing and viewing existing documents and websites, Qwen Deep Research focuses more on generating new research content from scratchby aggregating sources from the open web and presenting them in many ways.
The comparison suggests that while the overall concept of the two tools – AI-powered research – overlaps, they differ in their approach and target user experience.
Availability
Qwen Deep Research is now available live and via Qwen Chat application. This feature can be accessed using the following URL.
At the time of writing, no pricing details have been provided for Qwen3-Max or specific Deep Research capabilities.
What’s next for Qwen Deep’s research?
By combining research guidance, data evaluation, and content creation in multiple formats in one tool, Qwen Deep Research goals to streamline the path from idea to publishable results.
The integration of code, visuals, and voice makes it especially attractive to content creators, educators, and independent analysts who need to scale their research to web- or podcast-friendly forms without having to change platforms.
Still, comparisons to more specialized products like NotebookLM raise questions about how Qwen’s generalized approach stacks up in terms of depth, precision, and sophistication. Whether the power of multi-format outweighs these concerns may come all the way down to user priorities and whether or not they value one-click publishing over tight integration with existing notes and materials.
For now, Qwen signals that research doesn’t end with the document – it begins with it.
Let me know if you’d like this rebuilt into something shorter or tailored to a specific audience – newsletter, press style blog, internal team explainer, etc.