Arao AI
Back to search results

gemma-4-26B-A4B-it · Q4_K_S

Start a discussion
Apple M5 Max128GBvLLM v0.37L2 Cross-framework verified

This page aggregates 1 real-world runs of gemma-4-26B-A4B-it (Q4_K_S) on Apple M5 Max with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

136.41 tok/s

Decode

Decode speed

547.6 tok/s

Prefill

Prefill speed

2.29 s

TTFT

Time to first token

44.69 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

5 days ago

Last verified

Performance

  1. vLLM · Q3_K_S Decode 174.38 · Prefill 573.36 ·
  2. vLLM · Q4_K_S (current)Decode 136.41 · Prefill 547.6 ·
  3. Strata · Q2_K Decode 228.98 · Prefill 895.33 ·
  4. Strata · Q8_0 Decode 122.58 · Prefill 429.76 ·

Core figures

Decode (avg)
136.41 tok/s
Prefill (avg)
547.6 tok/s
TTFT (avg)
2.29 s
VRAM (avg)
44.69 GB
MTP acceptance rate
—
TTFB
1.09 GB
Power draw
59.3 W

Configuration

Member-level fields are taken from the most recent run

Model
gemma-4-26B-A4B-it
Quantization
Q4_K_S
Framework
vLLM
Version
v0.37
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
Apple M5 Max
Nominal VRAM
128 GB
Measured VRAM (avg)
44.69 GB
OS
macOS 15.0
Driver
Apple M5 Max
CUDA
N/A
Power draw
59.3 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-13

    v0.37 · macOS 15.0 · CUDA N/A · 2048 ctx

    136.41 tok/s

    Decode

    547.6 tok/s

    Prefill

    2.29 s

    TTFT

    44.69 GB

    VRAM

    —

    MTP

    59.3 W

    Power draw

    L2 batch seed 2026-10-13: vLLM on hw_id=28 running model_id=64 at Q4_K_S