DeepSeek-V4.1-Flash-1017 on one RTX 3090: Big updates to the Delta Inference Engine v0.93. Same 41 tok/s at 192GB/IQ2 at better quality, usable from 64 GB RAM, first token ~40% sooner, new Unsloth quants at full speed, EXL3, basic multi-GPU, public CI and more

1. 

   消息推送

   DeepSeek-V4.1-Flash-1017 on one RTX 3090: Big updates to the Delta Inference Engine v0.93. Same 41 tok/s at 192GB/IQ2 at better quality, usable from 64 GB RAM, first token ~40% sooner, new Unsloth quants at full speed, EXL3, basic multi-GPU, public CI and more

   7小时前

   DeepSeek-V4.1-Flash-1017 on one RTX 3090: Big updates to the Delta Inference Engine v0.93. Same 41 tok/s at 192GB/IQ2 at better quality, usable from 64 GB RAM, first token ~40% sooner, new Unsloth quants at full speed, EXL3, basic multi-GPU, public CI and more
   
   🌐 链接: https://www.reddit.com/r/LocalLLaMA/comments/1xoq4vq/deepseekv41flash1017_on_one_rtx_3090_big_updates/
   🏷️ 分组: Reddit｜AI·自托管·VPS
   🕒 时间: 2026-11-06 02:23:22

   Reddit

   

   From the LocalLLaMA community on Reddit: DeepSeek-V4.1-Flash-1017 on one RTX 3090: Big updates to the Delta Inference Engine v0.93.…

   Explore this post and more from the LocalLLaMA community

   7小时前