Arao AI
Back to search results

Llama-3.2-3B-Instruct · IQ3_XS

Start a discussion
NVIDIA H2097GBvLLM v0.31L2 Cross-framework verified

This page aggregates 1 real-world runs of Llama-3.2-3B-Instruct (IQ3_XS) on NVIDIA H20 with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

177.1 tok/s

Decode

Decode speed

886.15 tok/s

Prefill

Prefill speed

2.09 s

TTFT

Time to first token

6.01 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

16 days ago

Last verified

Performance

  1. llama.cpp · Q3_K_S Decode 171.35 · Prefill 761.41 ·
  2. vLLM · FP16 Decode 76.18 · Prefill 254.17 ·
  3. vLLM · IQ3_XS (current)Decode 177.1 · Prefill 886.15 ·

Core figures

Decode (avg)
177.1 tok/s
Prefill (avg)
886.15 tok/s
TTFT (avg)
2.09 s
VRAM (avg)
6.01 GB
MTP acceptance rate
—
TTFB
1.42 GB
Power draw
367.7 W

Configuration

Member-level fields are taken from the most recent run

Model
Llama-3.2-3B-Instruct
Quantization
IQ3_XS
Framework
vLLM
Version
v0.31
Context length
2048 tokens
Batch size
32
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 H20
Nominal VRAM
97 GB
Measured VRAM (avg)
6.01 GB
OS
Ubuntu 24.04
Driver
560.94
CUDA
12.6
Power draw
367.7 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

    v0.31 · Ubuntu 24.04 · CUDA 12.6 · 2048 ctx

    177.1 tok/s

    Decode

    886.15 tok/s

    Prefill

    2.09 s

    TTFT

    6.01 GB

    VRAM

    —

    MTP

    367.7 W

    Power draw

    L2 batch seed 2026-10-19: vLLM on hw_id=56 running model_id=70 at IQ3_XS