NVIDIA Releases NVFP4 Quantized Qwen3.8-Flash-Next on Hugging Face

DailyPapers, an automated curation account that tweets AI and ML papers from Hugging Face, posted that NVIDIA released the NVFP4 quantized Qwen3.8-Flash-Next. The post describes a 125B MoE model with hybrid attention now 63 percent smaller with minimal accuracy loss. It includes a link to an image showing a horizontal flowchart of five sequential stages of green hexagonal neural networks on a black background with green graphics. The account promotes submissions to the platform's papers section.