Arao AI
Back to search results

gemma-4-E2B-it · Q3_K_M

Start a discussion
AMD Ryzen AI Max+ 395128GBllama.cpp v0.21L2 Cross-framework verified

This page aggregates 1 real-world runs of gemma-4-E2B-it (Q3_K_M) on AMD Ryzen AI Max+ 395 with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

208.76 tok/s

Decode

Decode speed

678.14 tok/s

Prefill

Prefill speed

1.53 s

TTFT

Time to first token

4.96 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

2 days ago

Last verified

Performance

  1. llama.cpp · Q3_K_M (current)Decode 208.76 · Prefill 678.14 ·
  2. vLLM · IQ1_S Decode 291.55 · Prefill 1102.46 ·

Core figures

Decode (avg)
208.76 tok/s
Prefill (avg)
678.14 tok/s
TTFT (avg)
1.53 s
VRAM (avg)
4.96 GB
MTP acceptance rate
—
TTFB
1.67 GB
Power draw
60.9 W

Configuration

Member-level fields are taken from the most recent run

Model
gemma-4-E2B-it
Quantization
Q3_K_M
Framework
llama.cpp
Version
v0.21
Context length
32768 tokens
Batch size
256
GPU layers
20
Flash Attention
On

Hardware

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

GPU
AMD Ryzen AI Max+ 395
Nominal VRAM
128 GB
Measured VRAM (avg)
4.96 GB
OS
Windows 11 23H2
Driver
amdgpu-pro 24.20
CUDA
N/A
Power draw
60.9 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

    v0.21 · Windows 11 23H2 · CUDA N/A · 32768 ctx

    208.76 tok/s

    Decode

    678.14 tok/s

    Prefill

    1.53 s

    TTFT

    4.96 GB

    VRAM

    —

    MTP

    60.9 W

    Power draw

    L2 batch seed 2026-10-16: llama.cpp on hw_id=24 running model_id=66 at Q3_K_M