IT
Prenoti una chiamata

Local AI · Recipe · 4× DGX Spark

DeepSeek-V4.1-Flash FP8/MXFP4 with SparkRing on 4× DGX Spark

DeepSeek-V4.1-Flash FP8/MXFP4 with SparkRing on 4× DGX Spark: 57 tok/s according to github.com (dataset as of Sep 29, 2026).

by fujitsupolycom

57tok/sEveryday

1 request, decode at 16K context (profile table), thinking on, high

Source: github.com
Intelligence (original model) · Artificial Analysis40with thinkingno thinking 25
Instructions in the README Repo

Engine

Engine
vLLM (local-inference-lab Fork) via SparkRing (switchless NCCL ueber ConnectX-7)
Quantization
Stock-Checkpoint FP8/MXFP4 (DeepSeek)
Model family
DeepSeek-V4.1-Flash
Context
—
Parameters
763,2 GB Safetensors (510 GB on disk laut DB-Vorzeile)
Creator
fujitsupolycom
GitHub stars
91
Repo updated
Oct 6, 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)57 tok/scontext 16K · 1 Nutzer, je 16K Kontext; Thinking on, high Source 

Decode · parallel requests

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

  • 16 parallel requests275 tok/sPeak (decode) · context 16K · Thinking on, high Source 

Prefill by context

What you need

Hardware
4 × NVIDIA DGX Spark (GB10)
Engine
vLLM (local-inference-lab Fork) via SparkRing (switchless NCCL ueber ConnectX-7)

Notes

What matters before you rebuild it.

  • Custom kernel required

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
2× DGX Spark

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

by sfxnz

41.2tok/sEveryday

1 request, prose prompt (512 tokens), 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
  • Custom kernel required

EXL3 2.0 bpw MCG (routed experts, Branch 2.0bpw-mcg-viterbi-lmhead-mxfp8 @982b704), Rest FP8 (32x32-Blockscales), lm_head MXFP8, DSpark-Draft MXFP4TensorFold Fork sfxnz/TensorFold dsv41-recipe-engine4 @b86514a (v0.6.4 + deepseek_v41-Familie), TP=2, --parallel 4Weights Repo Updated Oct 8, 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