The official Qwen account on X posted on 6 November 2026 the first teaser for Alibaba's next generation of artificial intelligence models — widely read by the community as the Qwen4. The message contains just four words — "The becoming begins" — alongside an enigmatic poster. By 7 November, the post had already passed 2.2 million views and 17,000 likes.
The poster shows a flock of birds shaped like a fish over the sea at dusk, and carries the opening of a passage from the Zhuangzi, a Chinese text about 2,300 years old: "北冥有鱼 其名为_" — "in the northern sea there is a fish, and its name is ___", with the name left blank. The passage continues with Kun, the fish that grows until it turns into Peng, a giant bird. Reading the poster as a metaphor for transformation is the most common interpretation among commenters; Alibaba has not explained the image.
What is fact and what is interpretation: the teaser itself is a confirmed fact — the post exists and came from the official account. The link to Qwen4, however, is an inference: the post does not mention the name Qwen4 and gives no date, price, licence or specifications. The model has not been officially announced.
What is already known about Qwen4
What Alibaba has confirmed is that Qwen4 is in training. On 22 September, at the Apsara conference in Hangzhou, Qwen project lead Liu Dayiheng said the model used a new-generation architecture. On 29 October, at the European Apsara Conference in Amsterdam, a slide read "Qwen4 Series Coming Soon" and named four tiers — Max, Plus, Flash and 27B — and an attendee attributed a "next week" launch forecast to Alibaba Cloud CTO Feifei Li. No official channel has confirmed that date, which should be treated as a rumour.
What is still unconfirmed
As of 7 November, there was no Qwen4 repository in the Qwen organisation on Hugging Face, the platform where open weights are published, and no Qwen4 identifier in Alibaba Cloud's Model Studio. A preview of the architecture has existed since August: Qwen3.8-Flash-Next was presented as the model showing how Qwen4 will be built. The current flagship, Qwen3.8-Max, has 2.4 trillion parameters.
The Qwen family is one of the most used in the open-weights world: according to Alibaba, it has more than 3 billion downloads and over 300,000 derivative models. In China, Apple Intelligence uses Qwen models instead of ChatGPT.
Read also on Geekiki: Qwen3.8-Omni-Flash: Alibaba launches omni model with 1 million tokens.
Image credit: official Qwen teaser, posted by the @Alibaba_Qwen account on X on 6 November 2026 (reproduction: Pasquale Pillitteri).
Sources
- Pasquale Pillitteri — Qwen4 coming soon? Alibaba launches its official teaser
- PakTechStan — Alibaba Teases Qwen 4 With a Cryptic 'The Becoming Begins' Post
- Michael Leung — Qwen 4 Release: What Qwen's Fish-and-Birds Teaser Means
- Kingy AI — Alibaba Teaser Sparks Qwen 4 Speculation
Transparency: this article was produced with the assistance of artificial intelligence, cross-checked against public posts on X and internet sources. Please consult the original sources to confirm the information.
By Muse
