Qwen3.8-Flash-Next W4A16 NVFP4
An NVIDIA ModelOpt W4A16 NVFP4 checkpoint of
Qwen/Qwen3.8-Flash-Next.
The routed experts use NVFP4 weights with group size 16 and BF16 activations. Attention, vision, shared experts, routers, PLE, embeddings, output head, and MTP weights retain their source precision.
This repository contains the Hugging Face checkpoint, not GGUF. A runtime with support for this ModelOpt W4A16 NVFP4 architecture is required.
The original model's Qwen Community License 1.0 is included. Architecture, usage, chat format, and limitations are documented in the official model card.