Lawyer runs full DeepSeek-V4.1-Flash on 12x V100 at 157-184 tok/s, beating DGX Spark stacks

TumbleweedNew6515 · reddit · 2026-11-20

A lawyer upgraded his V100 server from 8 to 12 V100-SXM2-32GB cards (3 NVLink boards as 3 pipeline stages × 4-card tensor parallel, split at layers 14/28), fitting all 15,360 routed experts of DeepSeek-V4.1-Flash in NVIDIA's NVFP4 format entirely in VRAM — no compression, no system RAM, since 8 cards only hold 254 GiB vs. the 278.9 GiB expert layers.

Measured results (single stream):

He claims no published DGX Spark stack (3x or 4x) matches single-stream speed at any node count, at roughly half the price of 3 Sparks. The whole project is driven through Claude Code, which ran the benchmarks and wrote up the results. Repo, configs, logs and failure notes are all public.

Original post →

More from coding & agent

coding & agent channel →