Chinese e-commerce giant Alibaba’s famous and prolific Qwen team of AI researchers and modeling engineers has introduced a major expansion to its Qwen Deep Research tool, which is available as an optional modality that the user can activate in the web-based Qwen Chat (a ChatGPT competitor).
The update allows users to generate not only comprehensive research reports with well-organized citations, but also interactive web pages and multi-speaker podcasts, all with 1 or 2 clicks.
This functionality is part of a proprietary releasedistinct from many of Qwen’s previous open source model offerings.
While the feature is based on open source models Qwen3 encoder, Qwen Imageand Qwen3-TTS To leverage its core capabilities, the end-to-end experience, including research execution, web deployment, and audio generation, is hosted and operated by Qwen.
This means users benefit from an integrated and managed workflow without needing to configure infrastructure. That said, developers with access to open source models could, in theory, replicate similar functionality in private or commercial systems.
The update was announced through the team official. Account X (@Alibaba_Qwen) today October 21, 2025, expressing:
“Qwen Deep Research just received a major update. It now creates not only the report, but also a live web page and podcast, powered by Qwen3-Coder, Qwen-Image and Qwen3-TTS. Your insights, now visual and audible.”
Multi-format research results
The main workflow begins with a user request within the Qwen Chat interface. From there, Qwen collaborates by asking clarifying questions to shape the scope of the investigation, extracts data from the web and official sources, and analyzes or resolves any inconsistencies it finds, even generating custom code when necessary.
TO demo video posted by Qwen on X explains this process in Qwen Chat using the US SaaS market as an example.
In it, Qwen retrieves data from multiple industry sources, identifies discrepancies in market size estimates (for example, $206 billion vs. $253 billion), and highlights ambiguities in the United States’ share of the global figures. The wizard comments on differences in scope between sources and calculates a compound annual growth rate (CAGR) of 19.8% from 2020 to 2023, providing contextual analysis to support the raw numbers.
Once the investigation is complete, users can click the "eyeball" icon below the output result (see screenshot), which will display a PDF style report in the right pane.
Then, when viewing the report in the right pane, the user can click the "Create" in the upper right corner and select between the following two options:
-
"Web development" which produces a Live, professional-grade websiteautomatically implemented and Hosted by Qwenusing Qwen3-Coder for structure and Qwen-Image for visuals.
-
"Podcast," which, as it says, produces an audio podcastfeaturing dynamic multi-speaker narration generated by Qwen3-TTS, as well Hosted by Qwen for easy sharing and playback.
This allows users to quickly convert a single research project into multiple forms of content (written, visual and audible) with minimal additional input.
The website includes online 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 couldn’t find a way to preview the voice outputs before selecting them. It seems designed for deep listening on the go.
There was no way to change the output language I could see, so mine appeared in English, like my initial reports and messages, although Qwen’s LLMs are multimodal. The voices were a little more robotic than other AI tools I’ve used.
Here is an example of a web page I generated. on the commonalities of authoritarian regimes throughout history, another on UFO or UAP sightingsand below this paragraph, a podcast about UFO or UAP sightings.
While the website is hosted via a public link, the podcast must be downloaded by the user and cannot be linked publicly, from what I could tell from my brief usage so far.
Please note that the podcast is very different from the actual report: it is not just a read-through audio version, but rather a new format of two presenters discussing and joking about the topic using the report as a starting point.
The web page versions of the report also include new charts not found in the PDF report.
Comparisons to Google’s NotebookLM
While the new capabilities have been welcomed by many early adopters, comparisons have arisen with other research assistants, particularly Google’s. NotebookLMwhich recently came out of beta.
AI commentator and newsletter writer Chubby (@kimmonismus) noted in X:
“I’m really grateful that Qwen provides regular updates. That’s great.”
But the attempt to build a NotebookLM clone inside Qwen-3-max doesn’t look very promising compared to Google’s version.”
While NotebookLM is based on organizing and querying existing documents and web pages, Qwen Deep Research focuses more on Generate new research content from scratch.adding sources from the open web and presenting them in multiple modalities.
The comparison suggests that while the two tools overlap in overall concept (AI-assisted research), they diverge in focus and target user experience.
Availability
Qwen Deep Research is now live and available through Qwen Chat app. The function can be accessed with the following URL.
No pricing details have been provided for Qwen3-Max or Deep Research’s specific capabilities at the time of writing.
What’s next for Qwen Deep Research?
By combining research guidance, data analysis and multi-format content creation in a single tool, Qwen Deep Research aims to streamline the path from idea to publishable result.
The integration of code, images, and voice makes it especially attractive to content creators, educators, and independent analysts who want to expand their research to web or podcast formats without switching platforms.
Still, comparisons with more specialized offerings like NotebookLM raise questions about how Qwen’s generalized approach stacks up for depth, precision, and refinement. Whether the strength of its multi-format execution outweighs those concerns may depend on users’ priorities and whether they value one-click publishing rather than tight integration with existing notes and materials.
For now, Qwen is pointing out that the investigation doesn’t end with a document: it starts with one.
Let me know if you’d like to repackage it into something shorter or tailored to a particular audience: newsletter, press-style blog, internal team explainer, etc.
