Arao AI
Back to search results

Qwen3-8B · IQ3_XS

Start a discussion
NVIDIA RTX 309024GBvLLM v0.40L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3-8B (IQ3_XS) on NVIDIA RTX 3090 with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

108.64 tok/s

Decode

Decode speed

552.19 tok/s

Prefill

Prefill speed

0.14 s

TTFT

Time to first token

12.52 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

Today

Last verified

Performance

  1. vLLM · IQ3_XS (current)Decode 108.64 · Prefill 552.19 ·
  2. Ollama · IQ4_XS Decode 96.65 · Prefill 382.18 ·
  3. Ollama · Q8_0 Decode 48.02 · Prefill 169.17 ·
  4. Strata · Q5_K_S Decode 138.04 · Prefill 495.5 ·
  5. llama.cpp · Q8_0 Decode 84.9 · Prefill 636.8 ·
  6. llama.cpp · Q5_K_M Decode 123.7 · Prefill 927.8 ·
  7. llama.cpp · Q4_K_M Decode 149 · Prefill 1117.5 ·

Core figures

Decode (avg)
108.64 tok/s
Prefill (avg)
552.19 tok/s
TTFT (avg)
0.14 s
VRAM (avg)
12.52 GB
MTP acceptance rate
—
TTFB
0.87 GB
Power draw
273.8 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3-8B
Quantization
IQ3_XS
Framework
vLLM
Version
v0.40
Context length
16384 tokens
Batch size
8
GPU layers
999
Flash Attention
Off

Hardware

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

GPU
NVIDIA RTX 3090
Nominal VRAM
24 GB
Measured VRAM (avg)
12.52 GB
OS
Ubuntu 24.04
Driver
572.83
CUDA
12.4
Power draw
273.8 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-18

    v0.40 · Ubuntu 24.04 · CUDA 12.4 · 16384 ctx

    108.64 tok/s

    Decode

    552.19 tok/s

    Prefill

    0.14 s

    TTFT

    12.52 GB

    VRAM

    —

    MTP

    273.8 W

    Power draw

    L2 batch seed 2026-10-18: vLLM on hw_id=2 running model_id=53 at IQ3_XS