← Back to the wire

Qwen3.8-Flash-Next

AnnouncementModelAug 26, 2026

Qwen3.8-Flash-Next is a multimodal MoE model with 125B total and 6B active parameters, serving as an early preview of the architecture planned for Qwen4. The author tested Unsloth quantized versions on a DGX Spark, including the 72.5GB UD-IQ1_S and 78.9GB UD-Q2_K_XL builds, generating image outputs from each. The author's preferred result so far came from an xhigh reasoning effort run using UD-Q2_K_XL.

Receipt № 15891 source · awaiting confirmation ◐

Evidence

1source· awaiting independent confirmation

No score is assigned. Sources and their independence are shown in the citation chain below.

Citation chain · 1 source

OpenAICompanyHugging FaceCompanyUnslothCompanyQwen 3.8 27BModelQwen3.8-Flash-NextModelQwen4ModelUD-IQ1_SModelUD-Q2_K_XLModel
Canonical: https://simonwillison.net/2026/Aug/26/qwen38-flash-next/