Chinese e-commerce giant Alibaba's famously prolific Qwen team of AI model researchers and engineers has introduced a major expansion to its Qwen Deep Research tool, available as an optional modality that the user can activate within the web-based Qwen Chat (a competitor to ChatGPT).
With the update, users cannot only create comprehensive research reports with well-organized citations, but in addition interactive web pages and podcasts with multiple speakers – all inside 1-2 clicks.
This functionality is a component of a proprietary releaseunlike lots of Qwen's previous open source model offerings.
While the feature relies on the open source models Qwen3 coder, Qwen imageAnd Qwen3-TTS To drive its core functionality, the end-to-end experience – including conducting research, web delivery and audio generation – is very important hosted and operated by Qwen.
This means users profit from a managed, integrated workflow without having to configure the infrastructure. However, developers with access to the open source models could theoretically replicate similar functionality on private or business systems.
The update was announced via the team official X Account (@Alibaba_Qwen) today, October 21, 2025, with the statement:
“Qwen Deep Research has just received a serious upgrade. It now produces not only the report, but in addition a live website and podcast – supported by Qwen3-Coder, Qwen-Image and Qwen3-TTS. Your insights, now visual and audible.”
Research ends in multiple formats
The core workflow begins with a user request throughout the Qwen Chat interface. From there, Qwen works together by asking clarifying inquiries to shape the scope of the research, pulling data from the Internet and official sources, and analyzing or fixing any inconsistencies found – even creating custom code if vital.
A Demo video posted by Qwen on X carries out this process on Qwen Chat using the US SaaS market for example.
In it, Qwen pulls data from multiple industry sources, identifies discrepancies in market size estimates (e.g. $206 billion vs. $253 billion), and highlights ambiguities regarding the U.S. share of the worldwide numbers. Commenting on the differences in scale between sources, the wizard calculates a compound annual growth rate (CAGR) of 19.8% from 2020 to 2023 and provides contextual evaluation to support the raw numbers.
Once the research is complete, users can click on the “eyeball” icon below the output result (see screenshot), which can display a PDF-style report in the suitable pane.
Then, when the user views the report in the suitable pane, they will click the Create button in the highest right corner and choose one in all the next two options:
-
“Web developer” what creates a Live website at knowledgeable levelrobotically deployed and Hosted by Qwenwhere Qwen3-Coder is used for the structure and Qwen-Image for the visual representation.
-
“Podcast“which, because it is known as, produces an audio Podcastwith dynamic multi-speaker narration also generated by Qwen3-TTS Hosted by Qwen for straightforward sharing and playback.
This allows users to quickly transform a single research project into multiple types of content – ​​written, visual and audible – with minimal additional effort.
The website accommodates inline graphics generated by Qwen Image, making it suitable to be used in public presentations, classrooms or publishing.
The podcast feature allows users to make a choice from 17 different speaker names as host and seven as co-host. However, I couldn't discover a solution to preview the voiceovers before choosing them. It seems designed for intensive listening on the go.
There was no solution to change the voice output that I could see, so my output got here in English, as did my reports and initial prompts, regardless that the Qwen LLMs are multimodal. The voices were a bit more robotic than other AI tools I've used.
Here is an example of a web site I created on similarities between authoritarian regimes throughout history, one other about UFO or UAP sightingsand under this paragraph a podcast about UFO or UAP sightings.
While the web site is hosted via a public link, the podcast have to be downloaded by the user and can’t be publicly linked, as I even have seen from my temporary usage to date.
Note that the podcast may be very different from the actual report – not only a read-aloud audio version of it, but a brand new format by which two hosts discuss and banter concerning the topic, using the report as a place to begin.
The web page versions of the report also contain recent graphics not present in the PDF report.
Compare with Google's NotebookLM
While the brand new features have been well received by many early adopters, comparisons to other research assistants have emerged – particularly those from Google NotebookLMwhose beta version recently ended.
AI commentator and newsletter writer Chubby (@kimmonismus) listed on X:
“I'm really grateful that Qwen provides regular updates. That's great.
But attempting to create a NotebookLM clone inside Qwen-3-max doesn’t sound very promising in comparison with Google’s version.”
While NotebookLM relies on organizing and querying existing documents and web pages, Qwen Deep Research focuses more on that Generating recent research content from scratchAggregating sources from the open web and presenting them across multiple modalities.
The comparison suggests that while the 2 tools overlap in the overall concept – AI-powered research – they differ within the approach and goal of user experience.
Availability
Qwen Deep Research is now live and available via Qwen chat app. The function may be accessed with the next URL.
At this time, no pricing details have been provided for Qwen3-Max or the particular Deep Research features.
What’s next for Qwen Deep Research?
By combining research consulting, data evaluation and multi-format content creation in a single tool, Qwen Deep Research goals to streamline the journey from idea to publishable output.
The integration of code, images and language makes it particularly attractive for content creators, educators and independent analysts who wish to scale their research into web or podcast-friendly forms without switching platforms.
Still, comparisons with more specialized offerings like NotebookLM raise questions on how Qwen's general approach fares when it comes to depth, precision, and refinement. Whether the strength of multiformat execution outweighs these concerns may rely upon user priorities—and whether one-click publishing is more essential to them than tight integration with existing notes and materials.
For now, Qwen signals that research doesn’t end with a document, but begins with one.
Let me know when you'd prefer to repackage this into something shorter or tailored to a selected audience – newsletter, press-style blog, internal team statement, etc.
