Arao AI
Back to search results

gemma-4-E4B-it · IQ3_XS

Start a discussion
NVIDIA H100 80GB HBM380GBllama.cpp v0.46L2 Cross-framework verified

This page aggregates 1 real-world runs of gemma-4-E4B-it (IQ3_XS) on NVIDIA H100 80GB HBM3 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

297.43 tok/s

Decode

Decode speed

1,240.57 tok/s

Prefill

Prefill speed

1.94 s

TTFT

Time to first token

12.52 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

12 days ago

Last verified

Performance

  1. llama.cpp · IQ3_XS (current)Decode 297.43 · Prefill 1240.57 ·
  2. vLLM · Q3_K_S Decode 343.31 · Prefill 1851.05 ·

Core figures

Decode (avg)
297.43 tok/s
Prefill (avg)
1,240.57 tok/s
TTFT (avg)
1.94 s
VRAM (avg)
12.52 GB
MTP acceptance rate
—
TTFB
2.38 GB
Power draw
677.8 W

Configuration

Member-level fields are taken from the most recent run

Model
gemma-4-E4B-it
Quantization
IQ3_XS
Framework
llama.cpp
Version
v0.46
Context length
4096 tokens
Batch size
512
GPU layers
20
Flash Attention
Off

Hardware

Nominal and measured figures are shown side by side; whether it runs is the reader's call

GPU
NVIDIA H100 80GB HBM3
Nominal VRAM
80 GB
Measured VRAM (avg)
12.52 GB
OS
Ubuntu 24.04
Driver
565.57
CUDA
13.0
Power draw
677.8 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-24

    v0.46 · Ubuntu 24.04 · CUDA 13.0 · 4096 ctx

    297.43 tok/s

    Decode

    1,240.57 tok/s

    Prefill

    1.94 s

    TTFT

    12.52 GB

    VRAM

    —

    MTP

    677.8 W

    Power draw

    L2 batch seed 2026-10-24: llama.cpp on hw_id=55 running model_id=67 at IQ3_XS