Arao AI
Back to search results

Qwen3-8B · IQ2_XS

Start a discussion
NVIDIA RTX 309024GBvLLM v0.41L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3-8B (IQ2_XS) on NVIDIA RTX 3090 with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

113.39 tok/s

Decode

Decode speed

440.93 tok/s

Prefill

Prefill speed

2.12 s

TTFT

Time to first token

10.16 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

16 days ago

Last verified

Performance

  1. vLLM · IQ2_XS (current)Decode 113.39 · Prefill 440.93 ·
  2. Ollama · Q2_K Decode 98.87 · Prefill 381.75 ·
  3. vLLM · IQ3_XS Decode 108.64 · Prefill 552.19 ·
  4. Ollama · IQ4_XS Decode 96.65 · Prefill 382.18 ·
  5. Ollama · Q8_0 Decode 48.02 · Prefill 169.17 ·
  6. Strata · Q5_K_S Decode 138.04 · Prefill 495.5 ·
  7. llama.cpp · Q8_0 Decode 84.9 · Prefill 636.8 ·
  8. llama.cpp · Q5_K_M Decode 123.7 · Prefill 927.8 ·
  9. llama.cpp · Q4_K_M Decode 149 · Prefill 1117.5 ·

Core figures

Decode (avg)
113.39 tok/s
Prefill (avg)
440.93 tok/s
TTFT (avg)
2.12 s
VRAM (avg)
10.16 GB
MTP acceptance rate
—
TTFB
0.91 GB
Power draw
309 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3-8B
Quantization
IQ2_XS
Framework
vLLM
Version
v0.41
Context length
32768 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)
10.16 GB
OS
Windows 11 23H2
Driver
570.86
CUDA
12.8
Power draw
309 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

    v0.41 · Windows 11 23H2 · CUDA 12.8 · 32768 ctx

    113.39 tok/s

    Decode

    440.93 tok/s

    Prefill

    2.12 s

    TTFT

    10.16 GB

    VRAM

    —

    MTP

    309 W

    Power draw

    L2 batch seed 2026-10-19: vLLM on hw_id=2 running model_id=53 at IQ2_XS