Chinese language e-commerce big Alibaba’s well-known and prolific Qwen staff of AI mannequin researchers and engineers has launched a significant enlargement to its Qwen Deep Analysis device, which is on the market as an non-obligatory modality that the consumer can activate in web-based Qwen Chat (a competitor to ChatGPT).
The replace permits customers to generate not solely complete analysis experiences with well-organized citations, but additionally interactive internet pages and multi-speaker podcasts – all with 1-2 clicks.
This performance is a part of a proprietary modelnot like a lot of Qwen’s earlier open supply template choices.
Though the characteristic depends on open supply templates Qwen3 Encoder, Picture Qwenand Qwen3-TTS to energy its core capabilities, the end-to-end expertise—together with search execution, internet deployment, and audio technology—is hosted and operated by Qwen.
This implies customers profit from a managed and built-in workflow with out the necessity to configure infrastructure. That mentioned, builders with entry to open supply fashions may theoretically replicate comparable performance on personal or business methods.
The replace was introduced through the staff’s official web site Account X (@Alibaba_Qwen) immediately, October 21, 2025, stating:
“Qwen Deep Analysis simply received a significant replace. It now creates not simply the report, but additionally a reside webpage and podcast – powered by Qwen3-Coder, Qwen-Picture and Qwen3-TTS. Your insights, now visible and audible.”
Multi-format search outcomes
The primary workflow begins with a consumer request inside the Qwen Chat interface. From there, Qwen collaborates by asking clarifying inquiries to form the scope of the analysis, extracts knowledge from the online and official sources, and analyzes or resolves any inconsistencies he finds – even producing customized code when crucial.
ONE demo video posted by Qwen on X walks by means of this course of on Qwen Chat utilizing the US SaaS market for example.
In it, Qwen retrieves knowledge from a number of business sources, identifies discrepancies in market dimension estimates (e.g., $206 billion versus $253 billion), and highlights ambiguities within the U.S. share of worldwide numbers. The wizard feedback on variations in scope between sources and calculates a compound annual development charge (CAGR) of 19.8% from 2020 to 2023, offering contextual evaluation to help the uncooked numbers.
As soon as the survey is full, customers can click on the "eyeball" icon beneath the output end result (see screenshot), which can show a PDF-style report in the fitting pane.
Then, when viewing the report in the fitting pane, the consumer can click on the "To create" button within the higher proper nook and choose one of many following two choices:
-
"Net Developer" which produces a reside, professional-grade internet web pagerobotically deployed and hosted by Qwenutilizing Qwen3-Coder for construction and Qwen-Picture for visuals.
-
"Podcast," which, because it states, produces an audio podcastthat includes dynamic multi-speaker narration generated by Qwen3-TTS, additionally hosted by Qwen for simple sharing and playback.
This permits customers to shortly convert a single analysis mission into a number of types of content material – written, visible and audio – with minimal additional enter.
The location contains embedded graphics generated by Qwen Picture, making it appropriate to be used in public shows, lecture rooms or publications.
The podcast characteristic permits customers to pick from 17 completely different speaker names as host and seven as co-host, though I could not discover a strategy to preview the voice outputs earlier than choosing them. It appears designed for deep listening on the go.
There was no strategy to change the output language that I may see, so mine got here out in English, like my preliminary experiences and prompts, though Qwen LLMs are multimodal. The voices have been just a little extra robotic than different AI instruments I’ve used.
Right here is an instance of an online web page I generated about similarities in authoritarian regimes throughout history, another about UFO or UAP sightingsand beneath this paragraph, a podcast about UFO or UFO sightings.
Though the location is hosted through a public hyperlink, the podcast should be downloaded by the consumer and can’t be linked publicly, from what I’ve gathered from my temporary use up to now.
Observe that the podcast could be very completely different from the precise report – not only a straight-read audio model, however a brand new format of two hosts discussing and joking concerning the matter utilizing the report as a place to begin.
The online web page variations of the report additionally embody new charts not discovered within the PDF report.
Comparisons with Google’s NotebookLM
Whereas the brand new options have been well-received by many early adopters, comparisons to different search assistants have emerged, particularly Google’s. NotebookLMwhich lately got here out of beta.
AI Commentator and E-newsletter Author Gordinho (@kimmonismus) noted in the X:
“I am actually grateful that Qwen gives common updates. That is nice.
However the try and construct a NotebookLM clone inside Qwen-3-max would not look very promising in comparison with Google’s model.”
Whereas NotebookLM is designed to prepare and question current paperwork and internet pages, Qwen Deep Analysis focuses extra on producing new search content material from scratchaggregating sources from the open internet and presenting them in a number of modalities.
The comparability means that whereas the 2 instruments overlap within the common idea – AI-assisted search – they diverge within the method and purpose of the consumer expertise.
Availability
Qwen Deep Analysis is now reside and accessible through Qwen chat app. The useful resource could be accessed with the following URL.
No pricing particulars have been offered for Qwen3-Max or the precise Deep Analysis options as of this writing.
What’s subsequent for Qwen Deep Analysis?
By combining analysis steering, knowledge evaluation, and multi-format content material creation right into a single device, Qwen Deep Analysis goals to hurry the trail from thought to publishable end result.
The combination of code, visuals, and voice makes it particularly enticing to impartial content material creators, educators, and analysts who wish to scale their analysis to web- or podcast-friendly codecs with out switching platforms.
Nonetheless, comparisons with extra specialised choices like NotebookLM elevate questions on how Qwen’s generalized method stacks up in depth, accuracy, and refinement. Whether or not the energy of its multi-format execution outweighs these issues could also be a query of customers’ priorities – and whether or not they worth single-click publishing over tight integration with current notes and supplies.
For now, Qwen is signaling that analysis would not finish with a doc – it begins with one.
Let me know if you would like this repackaged into one thing shorter or tailor-made for a particular viewers – publication, press-style weblog, inner staff explainer, and so on.
