deepseek-v4.1IMAGE
First NVFP4 quant of DeepSeek V4.1 Flash 1017 lands on Hugging Face
Within seven hours of release, an NVFP4 quant appeared on Hugging Face, shrinking the model from 475 GiB to 400 GiB with bit-exact routed experts.
LibertAI 🕶️
@Libertai_DAIArchived 11 days agoOriginally posted 2026.10.17
LibertAI 🕶️ @Libertai_DAIOriginally posted 2026.10.17
DeepSeek dropped V4.1-Flash-1017 overnight. Less than 7 hours later it was on Hugging Face in NVFP4: 475 GiB → 400 GiB. The first NVFP4 of 1017. And the routed experts are bit-exact. 👇 🤗 https://huggingface.co/LibertAIDAI/DeepSeek-V4.1-Flash-1017-NVFP4
x.com/libertai_dai/status/2111480342158901748Archived 2026.10.18 16:45