Arao AI
Back to search results

gemma-4-31B-it · IQ2_XS

Start a discussion
NVIDIA RTX PRO 6000 Blackwell 96GB96GBvLLM v0.42L2 Cross-framework verified

This page aggregates 1 real-world runs of gemma-4-31B-it (IQ2_XS) on NVIDIA RTX PRO 6000 Blackwell 96GB with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

189.85 tok/s

Decode

Decode speed

737.75 tok/s

Prefill

Prefill speed

2.4 s

TTFT

Time to first token

33.35 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 · IQ2_XS (current)Decode 189.85 · Prefill 737.75 ·
  2. Ollama · IQ1_S Decode 184.42 · Prefill 842.53 ·
  3. Ollama · Q4_K_S Decode 134.3 · Prefill 614.37 ·

Core figures

Decode (avg)
189.85 tok/s
Prefill (avg)
737.75 tok/s
TTFT (avg)
2.4 s
VRAM (avg)
33.35 GB
MTP acceptance rate
—
TTFB
1.52 GB
Power draw
373 W

Configuration

Member-level fields are taken from the most recent run

Model
gemma-4-31B-it
Quantization
IQ2_XS
Framework
vLLM
Version
v0.42
Context length
4096 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 RTX PRO 6000 Blackwell 96GB
Nominal VRAM
96 GB
Measured VRAM (avg)
33.35 GB
OS
Ubuntu 22.04
Driver
572.83
CUDA
12.8
Power draw
373 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

    v0.42 · Ubuntu 22.04 · CUDA 12.8 · 4096 ctx

    189.85 tok/s

    Decode

    737.75 tok/s

    Prefill

    2.4 s

    TTFT

    33.35 GB

    VRAM

    —

    MTP

    373 W

    Power draw

    L2 batch seed 2026-10-19: vLLM on hw_id=68 running model_id=63 at IQ2_XS