Arao AI
Back to search results

Qwen3.8-9B · Q4_K_M

Start a discussion
NVIDIA RTX 409024GBStrata v0.31L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.8-9B (Q4_K_M) on NVIDIA RTX 4090 with Strata, contributed by 1 independent source platforms; metrics are averages of published measurements.

168.91 tok/s

Decode

Decode speed

879.31 tok/s

Prefill

Prefill speed

1.68 s

TTFT

Time to first token

17.6 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

17 days ago

Last verified

Performance

  1. vLLM · Q3_K_S Decode 144.28 · Prefill 508.23 ·
  2. Strata · Q4_K_M (current)Decode 168.91 · Prefill 879.31 ·
  3. vLLM · IQ4_XS Decode 131.27 · Prefill 643.79 ·
  4. vLLM · Q2_K Decode 152.53 · Prefill 504.65 ·
  5. Ollama · IQ3_XS Decode 121.23 · Prefill 396.17 ·

Core figures

Decode (avg)
168.91 tok/s
Prefill (avg)
879.31 tok/s
TTFT (avg)
1.68 s
VRAM (avg)
17.6 GB
MTP acceptance rate
0.72
TTFB
1.35 GB
Power draw
397.1 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.8-9B
Quantization
Q4_K_M
Framework
Strata
Version
v0.31
Context length
2048 tokens
Batch size
1024
GPU layers
12
Flash Attention
On

Hardware

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

GPU
NVIDIA RTX 4090
Nominal VRAM
24 GB
Measured VRAM (avg)
17.6 GB
OS
Windows 11 24H2
Driver
565.57
CUDA
12.8
Power draw
397.1 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-19

    v0.31 · Windows 11 24H2 · CUDA 12.8 · 2048 ctx

    168.91 tok/s

    Decode

    879.31 tok/s

    Prefill

    1.68 s

    TTFT

    17.6 GB

    VRAM

    0.72

    MTP

    397.1 W

    Power draw

    L2 batch seed 2026-10-19: Strata on hw_id=1 running model_id=52 at Q4_K_M