Arao AI
Back to search results

gemma-4-26B-A4B-it · IQ4_XS

Start a discussion
NVIDIA Jetson AGX Orin 64GB64GBvLLM v0.54L2 Cross-framework verified

This page aggregates 1 real-world runs of gemma-4-26B-A4B-it (IQ4_XS) on NVIDIA Jetson AGX Orin 64GB with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

42.78 tok/s

Decode

Decode speed

174.63 tok/s

Prefill

Prefill speed

2.05 s

TTFT

Time to first token

42.5 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

5 days ago

Last verified

Performance

  1. vLLM · IQ4_XS (current)Decode 42.78 · Prefill 174.63 ·
  2. vLLM · Q5_K_S Decode 34.07 · Prefill 117.53 ·
  3. Ollama · Q8_0 Decode 24.68 · Prefill 116.33 ·

Core figures

Decode (avg)
42.78 tok/s
Prefill (avg)
174.63 tok/s
TTFT (avg)
2.05 s
VRAM (avg)
42.5 GB
MTP acceptance rate
—
TTFB
0.34 GB
Power draw
15.5 W

Configuration

Member-level fields are taken from the most recent run

Model
gemma-4-26B-A4B-it
Quantization
IQ4_XS
Framework
vLLM
Version
v0.54
Context length
32768 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 Jetson AGX Orin 64GB
Nominal VRAM
64 GB
Measured VRAM (avg)
42.5 GB
OS
JetPack 6.0
Driver
NVIDIA L4T 36.3
CUDA
12.6
Power draw
15.5 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-13

    v0.54 · JetPack 6.0 · CUDA 12.6 · 32768 ctx

    42.78 tok/s

    Decode

    174.63 tok/s

    Prefill

    2.05 s

    TTFT

    42.5 GB

    VRAM

    —

    MTP

    15.5 W

    Power draw

    L2 batch seed 2026-10-13: vLLM on hw_id=76 running model_id=64 at IQ4_XS