When will Qwen 4 be released? The short answer is just below — and it comes straight from Alibaba’s own Apsara Conference announcement on 22 September 2026, not from rumour threads. I have been tracking every major model release this year for my SEO systems, so here is exactly what is confirmed, what is unconfirmed, and what I would do while you wait.
Short answer
- No release date exists: Alibaba’s official press release (22 September 2026) says Qwen 4 is currently in training, nothing more.
- The four tier names floating around — Qwen 4 Max, Flash, Plus and 27B — come from attendee reports, not the press release.
- Alibaba did confirm a roadmap: Qwen 4.5 and Qwen 5 are projected to scale to 5–10 trillion parameters.
- Apsara 2026 also brought AgentCore, the Qwen Intelligence phone platform and the Zhenwu V900 chip (Q1 2027).
- Practical move today: build on the current Qwen 3.8 line and keep your stack model-agnostic.
When will Qwen 4 be released? What Alibaba actually announced
Here is the confirmed position, straight from Alibaba’s official Apsara Conference press release (published 22 September 2026, carried on the Media OutReach newswire): Qwen 4 is currently in training. That is the entire official statement on timing. No launch date. No pricing. No downloadable weights. No API identifier.
The announcement came at Apsara 2026 — Alibaba Cloud’s annual flagship conference in Hangzhou, running under the theme “Intelligence Goes Beyond” — where the company laid out a full-stack AI roadmap covering new foundation models, proprietary AI chips, an agentic cloud platform and a mobile agent platform.
So if you have seen headlines implying Qwen 4 is out, or that a date has been set: it has not. What we have is an official in-training confirmation plus a roadmap for what comes after it.
| Claim | Status as of 23 September 2026 |
|---|---|
| Qwen 4 is in training | Confirmed in Alibaba’s official press release |
| Qwen 4 release date | Not announced |
| Qwen 4 pricing or API name | Not announced |
| Qwen 4 Max / Flash / Plus / 27B tiers | Attendee reports only — not in the press release |
| Qwen 4.5 and Qwen 5 at 5–10 trillion parameters | Stated as a forward-looking roadmap |
The Qwen 4.5 and Qwen 5 roadmap: 5 to 10 trillion parameters
The most striking line in the press release is not about Qwen 4 at all. Alibaba announced that the follow-up Qwen 4.5 and Qwen 5 model series are projected to scale up to 5 to 10 trillion parameters. A lot of coverage has mangled this into a Qwen 4 spec — it is not. It is a statement about the generations after Qwen 4, and it signals Alibaba intends to keep pushing frontier-scale training rather than only optimising smaller models.
For context on how fast this space moves, Grok, Kimi and the rest of the frontier pack have all shipped or teased major releases in the same window — my Grok 4.7 release date breakdown and Kimi K2.8 preview review cover the two I get asked about most.
🔥 Want this set up without the guesswork? If you are trying to work out which of these fast-moving models to actually build your SEO and content systems on, you do not need to guess. That is exactly what we do inside the AI Profit Boardroom — 3,700+ members, four live calls per week, daily tutorials, done-for-you templates and a 30-day roadmap.
Prefer to talk it through one-on-one first? Book a free SEO strategy session and we will map out your next steps together.
What Alibaba did release at Apsara 2026
While Qwen 4 stayed in the oven, the conference itself was far from empty. Per the same press release, Alibaba announced:
- AgentCore — a purpose-built agentic cloud platform for building and managing AI agents, plus an Agent Security Center aimed at enterprise deployments.
- Qwen Intelligence — a smartphone-focused agent platform pitched at next-generation AI phones.
- T-Head Zhenwu V900 — a new proprietary AI processor with 216 GB of GPU memory, scheduled for Q1 2027, which Alibaba says delivers three times the performance of its predecessor.
- Recursive self-improvement results — Alibaba reported that Qwen3.8-Max ran a month of fully automated training cycles (33 iterations) and improved its Artificial Analysis score from 40 to 45, and that a chip-design test saw over 60 hours of self-improvement reduce chip area by 42%.
⚠️ Vendor-reported numbers: The Qwen3.8-Max self-improvement scores (40 to 45), the 42% chip-area reduction and the “three times the performance” Zhenwu V900 figure all come from Alibaba’s own press release. They are vendor self-reported numbers and have not been independently verified.
What to do while you wait for Qwen 4
My honest advice: do not put your workflows on hold for an unannounced model. The current Qwen 3.8 line is strong, and the open models are free to run locally — my walkthrough on running Qwen3.8 27B with Ollama shows the exact setup I use. Build your content and agent pipelines on what is available today, keep the model layer swappable, and upgrading to Qwen 4 on day one becomes a config change rather than a rebuild.
That model-agnostic setup — pipelines that survive every new release cycle — is the core of what I teach inside the AI Profit Boardroom, and if you want a second pair of eyes on your own stack you can book a free SEO strategy session and we will go through it live.
When will Qwen 4 be released? The bottom line
Nobody outside Alibaba knows, and as of 23 September 2026 Alibaba itself has only confirmed that Qwen 4 is in training. Ignore any article that gives you a hard date — there is not one. The sensible play is to keep shipping with the current Qwen 3.8 generation, keep your pipelines model-agnostic, and check back here: I update this page the moment Alibaba publishes anything concrete.
FAQ: when will qwen 4 be released
Has Alibaba confirmed a Qwen 4 release date?
No. In the official Apsara Conference press release dated 22 September 2026, Alibaba said only that Qwen 4 is currently in training. No release date, pricing or downloadable weights have been announced.
What are Qwen 4 Max, Flash, Plus and 27B?
Those four tier names circulated in attendee reports and social media coverage of the Apsara keynote. They do not appear in the official press release I checked, so treat them as unconfirmed until Alibaba publishes model cards.
Will Qwen 4 be open weights?
Alibaba has not said. Previous Qwen releases have included open-weight models, but nothing about licensing or weights for Qwen 4 has been announced yet.
What is the Qwen 4.5 and Qwen 5 roadmap?
Alibaba’s press release says the Qwen 4.5 and Qwen 5 model series are projected to scale up to 5 to 10 trillion parameters. That figure is about future generations, not Qwen 4 itself.
What should I use while I wait for Qwen 4?
The current Qwen 3.8 line is the practical choice today, including the open models you can run locally through Ollama. My guide to Qwen3.8 27B on Ollama covers the setup.
Where will the Qwen 4 release be announced first?
Alibaba’s official Qwen channels and the Alibaba Cloud newsroom. I update this page as soon as anything concrete lands, so bookmark it if you want the confirmed answer rather than rumours.
Related reading
- Qwen3.8 27B on Ollama: full local setup
- Grok 4.7 release date: what is confirmed so far
- Kimi K2.8 preview review
Next steps: Do not wait for Qwen 4 to start building AI-powered SEO systems. Join 3,700+ members inside the AI Profit Boardroom for the live calls, daily tutorials and done-for-you templates, or book a free SEO strategy session and get a personal plan for your site.
About the author
Julian Goldie is an SEO agency owner with 10+ years in SEO, 394K+ subscribers on YouTube, a 100% job-success score on Upwork, 75K+ community members across his groups, and the author of a best-selling SEO book. He runs the AI Profit Boardroom community, and you can book a free SEO strategy session with his team any time.
Last updated September 2026. This is the living guide to when will qwen 4 be released — it gets updated as the tools change.
