Arao AI
Back to search results

gemma-4-31B-it · Q2_K

Start a discussion
Apple M4 Max128GBvLLM v0.42L2 Cross-framework verified

This page aggregates 1 real-world runs of gemma-4-31B-it (Q2_K) on Apple M4 Max with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

111.52 tok/s

Decode

Decode speed

400.99 tok/s

Prefill

Prefill speed

0.18 s

TTFT

Time to first token

37.91 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

Today

Last verified

Performance

  1. llama.cpp · Q3_K_M Decode 93.03 · Prefill 389.72 ·
  2. llama.cpp · Q8_0 Decode 53.15 · Prefill 271.21 ·
  3. vLLM · Q2_K (current)Decode 111.52 · Prefill 400.99 ·
  4. vLLM · Q8_0 Decode 54.33 · Prefill 226.48 ·

Core figures

Decode (avg)
111.52 tok/s
Prefill (avg)
400.99 tok/s
TTFT (avg)
0.18 s
VRAM (avg)
37.91 GB
MTP acceptance rate
—
TTFB
0.89 GB
Power draw
40.1 W

Configuration

Member-level fields are taken from the most recent run

Model
gemma-4-31B-it
Quantization
Q2_K
Framework
vLLM
Version
v0.42
Context length
2048 tokens
Batch size
8
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 M4 Max
Nominal VRAM
128 GB
Measured VRAM (avg)
37.91 GB
OS
macOS 15.0
Driver
Apple M4 Max
CUDA
N/A
Power draw
40.1 W

Sources & evidence

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

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

    v0.42 · macOS 15.0 · CUDA N/A · 2048 ctx

    111.52 tok/s

    Decode

    400.99 tok/s

    Prefill

    0.18 s

    TTFT

    37.91 GB

    VRAM

    —

    MTP

    40.1 W

    Power draw

    L2 batch seed 2026-10-10: vLLM on hw_id=26 running model_id=63 at Q2_K