Arao AI
Back to search results

gemma-4-E4B-it · Q5_K_M

Start a discussion
2x AMD Strix HalovLLM v0.42L2 Cross-framework verified

This page aggregates 1 real-world runs of gemma-4-E4B-it (Q5_K_M) on 2x AMD Strix Halo with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

149.23 tok/s

Decode

Decode speed

534.97 tok/s

Prefill

Prefill speed

0.6 s

TTFT

Time to first token

9.91 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 · IQ3_S Decode 197.35 · Prefill 823.81 ·
  2. vLLM · Q5_K_M (current)Decode 149.23 · Prefill 534.97 ·
  3. Strata · Q2_K Decode 311.82 · Prefill 1554.79 ·

Core figures

Decode (avg)
149.23 tok/s
Prefill (avg)
534.97 tok/s
TTFT (avg)
0.6 s
VRAM (avg)
9.91 GB
MTP acceptance rate
—
TTFB
1.19 GB
Power draw
60.6 W

Configuration

Member-level fields are taken from the most recent run

Model
gemma-4-E4B-it
Quantization
Q5_K_M
Framework
vLLM
Version
v0.42
Context length
16384 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
2x AMD Strix Halo
Measured VRAM (avg)
9.91 GB
OS
Ubuntu 24.04
Driver
amdgpu 6.10
CUDA
N/A
Power draw
60.6 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

    v0.42 · Ubuntu 24.04 · CUDA N/A · 16384 ctx

    149.23 tok/s

    Decode

    534.97 tok/s

    Prefill

    0.6 s

    TTFT

    9.91 GB

    VRAM

    —

    MTP

    60.6 W

    Power draw

    L2 batch seed 2026-10-16: vLLM on hw_id=88 running model_id=67 at Q5_K_M