FR
Réserver un appel

Local AI · Recipe · 1× Mac

Qwen3.8-Flash-Next 125B-A6B MLX 4/8-bit with mlx-serve on 1× Mac (M5 Ultra)

Qwen3.8-Flash-Next 125B-A6B MLX 4/8-bit with mlx-serve on 1× Mac (M5 Ultra): 155 tok/s according to github.com (dataset as of Sep 29, 2026).

by drowzeys

155tok/sEveryday

1 request, llmprobe standard prompt, thinking on, with MTP + PLD (per recipe)

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

Engine

Engine
mlx-serve 26.9.7-dev @933566c, MLX 0.32.3 (NAX an)
Quantization
MLX mixed 4/8-bit (ddalcu, 100 GiB), KV bf16
Model family
Qwen3.8-Flash-Next
Context
262,144
Parameters
125B MoE / 6B active
Creator
drowzeys
GitHub stars
1
Repo updated
Sep 30, 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)155 tok/sLlmprobe 0.6.13, greedy, Median aus 3 (143.6–165.8), 192 Token Source 
  • Peak (decode)277 tok/sVorhersagbarer Prompt (Spekulation, 7.84 tok/step); x4 Aggregat 166.8 Source 

Prefill by context

Time to first token

Shorter is better.

What you need

Hardware
1 × Mac / Apple Silicon
Engine
mlx-serve 26.9.7-dev @933566c, MLX 0.32.3 (NAX an)
Context
262,144 tokens

Notes

What matters before you rebuild it.

  • Experimental
  • License unclear

Sources

Related recipes

2× DGX Spark

Qwen3.8-Flash-Next 125B INT4-AutoRound with TensorFold on 2× DGX Spark (1M context)

by MiaAI-Lab

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 8, 2026

Details
2× DGX Spark

Qwen3.8-Flash-Next INT4-AutoRound (v5.2) with vLLM on 2× DGX Spark

by myllmbox

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 6, 2026

Details
2× DGX Spark

Qwen3.8-Flash-Next 125B NVFP4 with TensorFold on 2× DGX Spark (1M context)

by MiaAI-Lab

67.4tok/sEveryday

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

Intelligence40with thinking
  • Custom kernel required
  • License unclear

NVFP4 (nvidia, Experten) + FP8 n-gram/MTP, 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-DraftsWeights 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