Arao AI
Back to search results

Qwen3.8-9B · IQ3_S

Start a discussion
NVIDIA DGX Spark128GBvLLM v0.29L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.8-9B (IQ3_S) on NVIDIA DGX Spark with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

178.22 tok/s

Decode

Decode speed

972.1 tok/s

Prefill

Prefill speed

0.21 s

TTFT

Time to first token

14.57 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

2 days ago

Last verified

Performance

  1. vLLM · IQ3_S (current)Decode 178.22 · Prefill 972.1 ·
  2. vLLM · Q5_K_M Decode 122.43 · Prefill 399.31 ·
  3. Strata · Q3_K_S Decode 245.76 · Prefill 803.01 ·
  4. Ollama · Q8_0 Decode 88.03 · Prefill 479.67 ·

Core figures

Decode (avg)
178.22 tok/s
Prefill (avg)
972.1 tok/s
TTFT (avg)
0.21 s
VRAM (avg)
14.57 GB
MTP acceptance rate
—
TTFB
1.43 GB
Power draw
202.2 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.8-9B
Quantization
IQ3_S
Framework
vLLM
Version
v0.29
Context length
8192 tokens
Batch size
16
GPU layers
999
Flash Attention
On

Hardware

Nominal and measured figures are shown side by side; whether it runs is the reader's call

GPU
NVIDIA DGX Spark
Nominal VRAM
128 GB
Measured VRAM (avg)
14.57 GB
OS
macOS 15.0
Driver
555.85
CUDA
12.4
Power draw
202.2 W

Sources & evidence

1 measured records in total, each traceable to its original source

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-16

    v0.29 · macOS 15.0 · CUDA 12.4 · 8192 ctx

    178.22 tok/s

    Decode

    972.1 tok/s

    Prefill

    0.21 s

    TTFT

    14.57 GB

    VRAM

    —

    MTP

    202.2 W

    Power draw

    L2 batch seed 2026-10-16: vLLM on hw_id=54 running model_id=52 at IQ3_S