Local AI · Recipe · 2× DGX Spark

Qwen3.8-Flash-Next NVFP4 with vLLM on 2× DGX Spark (vision)

Qwen3.8-Flash-Next NVFP4 with vLLM on 2× DGX Spark (vision): 43.2 tok/s according to github.com (dataset as of Nov 3, 2026).

by sfxnz

43.2tok/sEveryday

1 request, prose prompt, with MTP (per recipe)

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

Engine

Engine
vLLM
Quantization
ModelOpt NVFP4 W4A4 routed experts, PLE-n-gram-Tabelle per-Tensor FP8, MTP-Experten FP8_BLOCK_SCALES (gs 128), KV auto, Marlin/FlashInfer via moe-backend auto
Model family
Qwen3.8-Flash-Next
Context
262,144
Parameters
HF-API safetensors total 119,602,003,859
Creator
sfxnz
GitHub stars
1
Repo updated
Oct 1, 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)43.2 tok/sBench_decode.py, streamed greedy, thinking off, 200 Completion-Tokens, 3-run-Median; evidence opt-c4-native-ctx-262144: 43.16; TTFT p50 0.17 s Source 
  • Peak (decode)55.5 tok/sStructured (Count 1→200); evidence: 55.47; TTFT p50 0.15 s Source 

What you need

Hardware
2 × NVIDIA DGX Spark (GB10)
Weights
Engine
vLLM
Context
262,144 tokens

Related recipes

2× DGX Spark
89.3tok/sEveryday

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

Intelligenceno independent value
  • Custom kernel required
  • License unclear

INT4-AutoRound (GPTQ int4, Top-5-Experten) + FP8 n-gram-Tabelle, FP8 KVTensorFold tensorfold-native (Zig, zig-flashnext db281878 + patches), Image ghcr.io/miaai-lab/qwen3.8-flash-dual-dgx-sparks-tensorfold:zig-db28187-f3e1300f4f62, TP=2 über CX7/RoCE, MTP + Copy-Drafts (QUANT=int4ar)Weights Repo Updated Oct 26, 2026

Details
2× DGX Spark
74tok/sEveryday

1 request, prose prompt (long story), thinking off, with MTP Source 

Intelligence40with thinking
  • Custom kernel required
  • License unclear

INT4-AutoRound (azampatti): int4-Experten, fp8-Seitenschichten; n-gram-Tabelle beim Laden fp8 → NVFP4, halb pro Box auf der GPU; KV-Pin 29G/Box (1.829.182 Tokens)vLLM 0.30 (Image myllmbox/qwen38-flash-next-cluster-vllm:v5.2; RecoverSSM, dynamischer Draft-Depth 3–7 in Full-CUDA-Graphs, sampled Drafts + Block-Verifikation, One-Shot-RoCE-All-Reduce, Dual-Rail, TP=2, 64 Seats) — Kit v5.2Weights Repo Updated Oct 29, 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.