---
format: "aidr-story-markdown/v1"
id: "6f7afceb21266288d963b5497add1799f473a3126edcba0b4c2fca860e0feae3"
canonical_url: "https://aidr.today/6f7afceb?lang=en"
title: "Alibaba Ships 120B Qwen3.8-Flash-Next, First Glimpse of Qwen4 Architecture"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-08-25T14:52:31.000Z"
category: "Models"
topics: ["alibaba","qwen","open-source","moe"]
source_urls: ["https://huggingnews.com/ai/alibaba-ships-120b-qwen38-flash-next-first-glimpse-of-qwen4-architecture-8dc11599","https://x.com/bdsqlsz/status/2092210503498309862","https://x.com/kimmonismus/status/2092215963085111547","https://x.com/jun_song/status/2092217898752483570","https://x.com/danielhanchen/status/2092222459550585019","https://x.com/TheZachMueller/status/2092232217074295065","https://x.com/Alibaba_Qwen/status/2092489636287570169","https://x.com/kimmonismus/status/2092541014183461369"]
summary: "The multimodal Mixture-of-Experts (MoE) system will be available for open-weight download starting tomorrow, August 26. Alibaba designed the model with 120 billion total parameters and 6 billion active parameters to showcase a next-generation framework. This release provides developers early access to the structural improvements that will underpin the upcoming Qwen4 family, marking the first public look at the new model generation. UnslothAI is preparing day zero support for the release to facilitate immediate integration for users. The shift toward a 120B parameter scale follows the previous Qwen-3.8 27b version and signals a move toward larger capacities for open-weight multimodal models."
---

# Alibaba Ships 120B Qwen3\.8\-Flash\-Next, First Glimpse of Qwen4 Architecture

> [Open the canonical story](<https://aidr.today/6f7afceb?lang=en>)

**Published:** 2026-08-25T14:52:31.000Z
**Category:** Models
**Topics:** alibaba, qwen, open\-source, moe

## Summary

The multimodal Mixture\-of\-Experts \(MoE\) system will be available for open\-weight download starting tomorrow, August 26\. Alibaba designed the model with 120 billion total parameters and 6 billion active parameters to showcase a next\-generation framework\. This release provides developers early access to the structural improvements that will underpin the upcoming Qwen4 family, marking the first public look at the new model generation\. UnslothAI is preparing day zero support for the release to facilitate immediate integration for users\. The shift toward a 120B parameter scale follows the previous Qwen\-3\.8 27b version and signals a move toward larger capacities for open\-weight multimodal models\.

## Sources

- [Story source](<https://huggingnews.com/ai/alibaba-ships-120b-qwen38-flash-next-first-glimpse-of-qwen4-architecture-8dc11599>)
- [Story source](<https://x.com/bdsqlsz/status/2092210503498309862>)
- [Supporting source](<https://x.com/kimmonismus/status/2092215963085111547>)
- [Supporting source](<https://x.com/jun_song/status/2092217898752483570>)
- [Supporting source](<https://x.com/danielhanchen/status/2092222459550585019>)
- [Supporting source](<https://x.com/TheZachMueller/status/2092232217074295065>)
- [Story source](<https://x.com/Alibaba_Qwen/status/2092489636287570169>)
- [Supporting source](<https://x.com/kimmonismus/status/2092541014183461369>)