Arao AI
Back to search results

gemma-4-12B-it · Q4_K_M

Start a discussion
Apple M3 Max36GBllama.cpp v0.53L2 Cross-framework verified

This page aggregates 1 real-world runs of gemma-4-12B-it (Q4_K_M) on Apple M3 Max with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

85.7 tok/s

Decode

Decode speed

345.14 tok/s

Prefill

Prefill speed

1.16 s

TTFT

Time to first token

22.76 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

Today

Last verified

Performance

  1. llama.cpp · Q4_K_M (current)Decode 85.7 · Prefill 345.14 ·
  2. llama.cpp · Q5_K_S Decode 77.34 · Prefill 357.59 ·
  3. Ollama · IQ3_S Decode 94.58 · Prefill 396.89 ·

Core figures

Decode (avg)
85.7 tok/s
Prefill (avg)
345.14 tok/s
TTFT (avg)
1.16 s
VRAM (avg)
22.76 GB
MTP acceptance rate
—
TTFB
0.69 GB
Power draw
65.2 W

Configuration

Member-level fields are taken from the most recent run

Model
gemma-4-12B-it
Quantization
Q4_K_M
Framework
llama.cpp
Version
v0.53
Context length
16384 tokens
Batch size
512
GPU layers
99
Flash Attention
Off

Hardware

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

GPU
Apple M3 Max
Nominal VRAM
36 GB
Measured VRAM (avg)
22.76 GB
OS
macOS 15.0
Driver
Apple M3 Max
CUDA
N/A
Power draw
65.2 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

    v0.53 · macOS 15.0 · CUDA N/A · 16384 ctx

    85.7 tok/s

    Decode

    345.14 tok/s

    Prefill

    1.16 s

    TTFT

    22.76 GB

    VRAM

    —

    MTP

    65.2 W

    Power draw

    L2 batch seed 2026-10-10: llama.cpp on hw_id=4 running model_id=65 at Q4_K_M