deepseek-v4.1IMAGE

单张 RTX 5090 跑 V4.1 Flash 274GB 模型：双引擎对比

作者用一张 RTX 5090 加 121GB 内存跑 256 专家的 DeepSeek V4.1 Flash，达到两位数 tok/s，并对比两个引擎，结论是瓶颈在内存分层放置。

Lyes Sbahi

@lyesbahi5 天前收录原推发布于 2026.10.25

Lyes Sbahi @lyesbahi原推发布于 2026.10.25

DeepSeek-V4.1-Flash (256 experts, 274 GB) running at double-digit tokens/s on one RTX 5090 and 121 GB of RAM was unthinkable a year ago. Two engines do it today. We benched both - and the lesson is about RAM placement, not about anyone losing

x.com/lyesbahi/status/2114091823698158074采集于 2026.10.25 12:02