DeepSeek-4.1-Flash Hits ~676 tok/s Locally on 4× RTX PRO 6000

Users report running raw DeepSeek-V4.1-Flash-1017 weights locally on 4× RTX PRO 6000 GPUs (384GB VRAM), achieving single-user decode speeds of about 670–676 tok/s in tests including code generation.

2026-10-22 ~ 2026-10-23 · 2 related posts

• DeepSeek-V4.1-Flash hits 675.9 tok/s decode on 4× RTX PRO 6000 workstation — 0xSero · 2026-10-22
• Local Run of Suspected DeepSeek-4.1-Flash Hits 670 TPS on 4x RTX Pro 6000 — Kyrannio · 2026-10-23