EN
Book a call

Local AI · Recipe · 8× DGX Spark

DeepSeek-V4.1-Flash FP8/MXFP4 with SparkRing on 8× DGX Spark

DeepSeek-V4.1-Flash FP8/MXFP4 with SparkRing on 8× DGX Spark: 74 tok/s according to github.com (dataset as of Nov 3, 2026).

by fujitsupolycom

74tok/sEveryday

1 request, generation at 32K context (speed table, ring of eight), thinking on, high, with DSpark (per recipe); experimental profile

Source: github.com
Intelligence (original model) · Artificial Analysis40with thinkingno thinking 25

Install: DeepSeek-V4.1-Flash FP8/MXFP4 with SparkRing on 8× DGX Spark

Command
sudo sparkring install --profile deepseek-v41-flash-tp8

Full instructions in the README Repo

Engine

Engine
vLLM (local-inference-lab Fork) via SparkRing (SIRCL-Collectives, switchless RDMA ueber ConnectX-7), Image Release 2026.10.2
Quantization
Stock-Checkpoint FP8/MXFP4 (DeepSeek)
Model family
DeepSeek-V4.1-Flash
Context
1,048,576
Parameters
384 Routed-Experten (MoE); 510,3 GB Safetensors (475,3 GiB)
Creator
fujitsupolycom
GitHub stars
102
Repo updated
Nov 2, 2026

Measurements

Every sourced value of this recipe, each with its condition and source. Bars relative to the largest value in the group.

Decode · 1 request

  • Everyday (decode)74 tok/scontext 32K · 1 Stream, 32K Kontext, Ring aus acht Sparks (TP8, nur SIRCL-Ring-Sessions, NCCL aus), Release-Image 2026.10.2, GPU-Takt fest 2.418 MHz, ein Start; Thinking on, high; DSpark mit 5 Tokens Source 

Decode · parallel requests

Sums over all streams are marked; per request it is correspondingly less.

  • sum over 8 streams259 tok/sPeak (decode) · context 32K · Aggregate of 8 streams at 32K context, ring of eight; thinking on, high Source 

Prefill by context

  • 32K4,851 tok/sEin Prompt mit ca. 32.200 Tokens (Client-Messung bis zum ersten Token) Source 

What you need

Hardware
8 × NVIDIA DGX Spark (GB10)
Engine
vLLM (local-inference-lab Fork) via SparkRing (SIRCL-Collectives, switchless RDMA ueber ConnectX-7), Image Release 2026.10.2
Context
1,048,576 tokens

Notes

What matters before you rebuild it.

  • Custom kernel required
  • Experimental

Sources

Related recipes

2× DGX Spark

DeepSeek-V4.1-Flash EXL3 2.0 bpw with vLLM on 2× DGX Spark

by sfxnz

42.6tok/sEveryday

1 request, prose prompt, with DSpark (per recipe) Source 

Intelligence40with thinkingno thinking 25Original model's value. Heavily compressed (~2 bit), intelligence is likely much lower.
  • ≤ 3 bit: quantization may cost quality
  • Experimental
  • Custom kernel required

EXL3 2.0 bpw MCG (routed experts, tail-biting Viterbi re-encoded, scale refit), lm_head MXFP8, KV fp8 (8-GiB-Pin), DSpark-3vLLMWeights Repo Updated Sep 27, 2026

Details

← Back to overview

Local AI in your company?

In a workshop we work out which models and which hardware fit your tasks, and build the first agent on your infrastructure.

See the AI workshopBack to overview
Cookies

With your consent we use Google Analytics (Google, USA) and our own audience measurement to understand how our website is used. Strictly necessary storage, such as your cookie choice, happens without consent. You can change your choice at any time via “Cookie settings” in the footer. Privacy Policy · Cookie Policy