Arao AI
Back to search results

Qwen2.5-Omni-3B · IQ1_S

Start a discussion
NVIDIA Jetson AGX Orin 64GB64GBvLLM v0.46L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen2.5-Omni-3B (IQ1_S) on NVIDIA Jetson AGX Orin 64GB with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

135.17 tok/s

Decode

Decode speed

565.39 tok/s

Prefill

Prefill speed

1.13 s

TTFT

Time to first token

4.75 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

2 days ago

Last verified

Performance

  1. vLLM · IQ1_S (current)Decode 135.17 · Prefill 565.39 ·
  2. vLLM · Q2_K Decode 118.95 · Prefill 465.31 ·
  3. Strata · Q3_K_S Decode 147.6 · Prefill 732.7 ·
  4. llama.cpp · Q2_K Decode 113.33 · Prefill 517.85 ·
  5. llama.cpp · Q3_K_L Decode 91.94 · Prefill 356.4 ·
  6. vLLM · Q4_K_S Decode 88.52 · Prefill 343.83 ·

Core figures

Decode (avg)
135.17 tok/s
Prefill (avg)
565.39 tok/s
TTFT (avg)
1.13 s
VRAM (avg)
4.75 GB
MTP acceptance rate
—
TTFB
1.08 GB
Power draw
58.8 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen2.5-Omni-3B
Quantization
IQ1_S
Framework
vLLM
Version
v0.46
Context length
32768 tokens
Batch size
16
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 Jetson AGX Orin 64GB
Nominal VRAM
64 GB
Measured VRAM (avg)
4.75 GB
OS
JetPack 6.0
Driver
NVIDIA L4T 36.3
CUDA
12.6
Power draw
58.8 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-16

    v0.46 · JetPack 6.0 · CUDA 12.6 · 32768 ctx

    135.17 tok/s

    Decode

    565.39 tok/s

    Prefill

    1.13 s

    TTFT

    4.75 GB

    VRAM

    —

    MTP

    58.8 W

    Power draw

    L2 batch seed 2026-10-16: vLLM on hw_id=76 running model_id=60 at IQ1_S