Local AI · Recipe · 1× DGX Spark

DeepSeek-V4-Flash 180B NVFP4 with vLLM on 1× DGX Spark

DeepSeek-V4-Flash 180B NVFP4 with vLLM on 1× DGX Spark: 24.4 tok/s according to huggingface.co (dataset as of Nov 3, 2026).

by 0xSero

24.4tok/sPeak

1 request, best value reported by source

Source: huggingface.co
Intelligence (original model) · Artificial Analysisno independent value

Engine

Engine
vLLM
Quantization
MXFP4/NVFP4-style experts, FP8 MLA KV
Model family
DeepSeek-V4-Flash
Context
200,000
Parameters
180B (REAP K160 aus 641B laut Card)
Creator
0xSero
GitHub stars
25
Repo updated
Sep 17, 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

  • Peak (decode)24.4 tok/sMTP2; Kontext des Runs laut Card nicht eindeutig Source 

Prefill by context

What you need

Hardware
1 × NVIDIA DGX Spark (GB10)
Engine
vLLM
Context
200,000 tokens

Related recipes

← 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.