Arao AI
Back to search results

Qwen3.8-9B · Q3_K_S

Start a discussion
NVIDIA RTX 509032GBllama.cpp v0.30L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.8-9B (Q3_K_S) on NVIDIA RTX 5090 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

289.63 tok/s

Decode

Decode speed

1,418.95 tok/s

Prefill

Prefill speed

1.71 s

TTFT

Time to first token

14.38 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

17 days ago

Last verified

Performance

  1. llama.cpp · IQ2_XS Decode 304.42 · Prefill 1543.97 ·
  2. llama.cpp · Q3_K_S (current)Decode 289.63 · Prefill 1418.95 ·
  3. vLLM · Q2_K Decode 312.49 · Prefill 1617.22 ·
  4. vLLM · Q8_0 Decode 156.33 · Prefill 685.96 ·
  5. Ollama · Q3_K_S Decode 237.4 · Prefill 1150.93 ·
  6. Strata · Q3_K_L Decode 349.94 · Prefill 1124.59 ·
  7. vLLM · Q3_K_S Decode 286.31 · Prefill 1535.26 ·
  8. Strata · IQ1_S Decode 313.75 · Prefill 1522.32 ·
  9. Ollama · IQ4_XS Decode 224.73 · Prefill 806.05 ·
  10. Ollama · Q5_K_S Decode 182.89 · Prefill 696.63 ·
  11. Strata · Q5_K_S Decode 276.86 · Prefill 910.66 ·

Core figures

Decode (avg)
289.63 tok/s
Prefill (avg)
1,418.95 tok/s
TTFT (avg)
1.71 s
VRAM (avg)
14.38 GB
MTP acceptance rate
—
TTFB
2.32 GB
Power draw
396.4 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.8-9B
Quantization
Q3_K_S
Framework
llama.cpp
Version
v0.30
Context length
32768 tokens
Batch size
512
GPU layers
99
Flash Attention
On

Hardware

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

GPU
NVIDIA RTX 5090
Nominal VRAM
32 GB
Measured VRAM (avg)
14.38 GB
OS
Ubuntu 22.04
Driver
572.83
CUDA
12.6
Power draw
396.4 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

    v0.30 · Ubuntu 22.04 · CUDA 12.6 · 32768 ctx

    289.63 tok/s

    Decode

    1,418.95 tok/s

    Prefill

    1.71 s

    TTFT

    14.38 GB

    VRAM

    —

    MTP

    396.4 W

    Power draw

    L2 batch seed 2026-10-19: llama.cpp on hw_id=13 running model_id=52 at Q3_K_S