Alibaba Open-Sources Qwen3.8-Flash-Next: 125B Ultra-Sparse MoE Previewing Qwen4 Architecture
On August 26, Alibaba's Qwen team released and open-sourced the multimodal MoE model Qwen3.8-Flash-Next on Hugging Face, positioned as an early preview of the Qwen4 architecture. The community quickly launched discussions around quantization, fine-tuning, inference deployment, and benchmarks, making it one of the most-watched open-source model events of the day.
Confirmed
- The official accounts (@AlibabaQwen, @千问大模型) confirmed the model uses a multimodal MoE architecture with 125B total parameters and only 6B active per token, targeting extreme cost-efficiency — @千问大模型 cited a 9x cost reduction.
- Architecture upgrades include GDN+QSA hybrid attention, GatedResidual multi-branch residuals, and N-gram Embedding for expanded capacity (an additional 51B n-gram embedding parameters).
- The model supports image-text-to-text multimodal input, is compatible with the safetensors format and API, and an FP8 version is open-sourced on Hugging Face.
- Several community members (@kimmonismus, @SimplyAnnisa) relayed benchmark results showing strong performance — @SimplyAnnisa reported wins in 8 of 9 comparison benchmarks, and @kimmonismus claimed it beat Claude Opus.
- @sammcj opened a centralized discussion thread covering quantized variants, fine-tuning and de-tuning, chat templates, inference server configs, and model comparisons.
Unconfirmed
- @ariG23498 stated the parameter count is 180B, inconsistent with the official 125B total + 51B n-gram embedding figure — possibly an alternative counting that sums both; the official release remains the source of truth.
- Claims such as "6B active parameters beating Claude Opus" come from community relays; exact test conditions and official leaderboard details await the full technical report.
Why it matters
- As the first preview of the Qwen4 architecture, its ultra-sparse MoE design (6B/125B activation ratio) and new components like GDN+QSA signal the technical direction of Alibaba's next-generation models.
- The drastically reduced active parameters cut inference costs significantly; if the benchmark results hold up, it poses a direct cost-performance challenge to high-end closed-source models like Claude Opus.
2026-08-26 ~ 2026-08-26 · 11 related posts
- Episode 1: Alibaba Announces Qwen3.8-Flash-Next, a Qwen4 Architecture Preview Set for August 26 Release(2026-08-25, 13 posts)
- Episode 2: Alibaba Open-Sources Qwen3.8-Flash-Next: 125B Ultra-Sparse MoE Previewing Qwen4 Architecture(2026-08-26, 11 posts)
Primary sources
- Qwen 3.8 Flash Next Release: Megathread for Quants, Fine-Tunes, and Benchmarks — sammcj · 2026-08-26
- [source] Alibaba Unveils Qwen3.8-Flash Preview: MoE Model with Next-Gen Architecture — Alibaba_Qwen · 2026-08-26
- [source] Alibaba Qwen Releases Qwen3.8-Flash: 6B Active Params, 9x Cost Reduction — 千问大模型 · 2026-08-26
- Qwen 3.8 Flash-Next: 6B-Active Open Model Beats Claude Opus 4.6 Max — kimmonismus · 2026-08-26
- Alibaba Releases 180B Qwen3.8-Flash-Next Multimodal Model — ariG23498 · 2026-08-26
- Alibaba Releases Multimodal Model Qwen3.8-Flash-Next on Hugging Face — Qwen · 2026-08-26
- [source] Qwen3.8-Flash-Next architecture: 125B params, sparse MoE, 262K context — Alibaba_Qwen · 2026-08-26
- Alibaba Releases Qwen3.8-Flash: Insane Price-Performance, Hints at Qwen4 Arch — jiqizhixin · 2026-08-26
- Qwen 3.8 Flash Next Released: Beats DS V4 Flash with Half the Parameters — elemental-mind · 2026-08-26
- Alibaba's Qwen3.8 with 51B N-gram embeddings now available on SGLang — Alibaba_Qwen · 2026-08-26
1 near-duplicate retellings: SimplyAnnisa