NVIDIA releases NVFP4 quantized DeepSeek-V4.1-Flash, 665B params on HF
TheZachMueller · x · 2026-10-31
NVIDIA published DeepSeek-V4.1-Flash-1017-NVFP4 on Hugging Face: a 665B-parameter FP4/8-bit quantized version (tensor types include BF16/F32/F8E4M3/U8/I8), MIT-licensed, produced with Model Optimizer, supporting image-text-to-text. It's NVIDIA's official low-precision build of DeepSeek V4.1 aimed at cheaper inference deployment.
Related event: NVIDIA releases NVFP4 quantized DeepSeek-V4.1 Flash model(2 posts)→
More from Infra
- How Different NVIDIA GPUs Change Model Outputs via BF16, FP8, FP4 Precision — Kyrannio · 2026-10-31
- Kubernetes AI Conformance becomes a permanent subproject under SIG Architecture — TerryTangYuan · 2026-10-31
- Google shares first AI-generated response from space — a haiku, per Pichai — LinkedInNews · 2026-10-31
- Debugging llama.cpp: sliding-window context checkpoints silently eat your RAM — jomi-se · 2026-10-31
- Cornell Professor's Startup Makora Wins $1.75M to Make AI Faster and Cheaper to Run — srush_nlp · 2026-10-31
- Transformers.js monthly npm downloads grow 9x in a year, from 1.64M to ~15M — nicodotdev · 2026-10-31