Arao AI
Back to search results

mammoth-1 · Q5_K_M

Start a discussion
Apple M2 Ultrallama.cpp v0.39L2 Cross-framework verified

This page aggregates 1 real-world runs of mammoth-1 (Q5_K_M) on Apple M2 Ultra with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

67.37 tok/s

Decode

Decode speed

314.26 tok/s

Prefill

Prefill speed

2.02 s

TTFT

Time to first token

54.83 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

13 days ago

Last verified

Performance

  1. llama.cpp · IQ2_M Decode 102.29 · Prefill 417.96 ·
  2. llama.cpp · Q5_K_M (current)Decode 67.37 · Prefill 314.26 ·
  3. Ollama · IQ1_S Decode 100.6 · Prefill 448.02 ·

Core figures

Decode (avg)
67.37 tok/s
Prefill (avg)
314.26 tok/s
TTFT (avg)
2.02 s
VRAM (avg)
54.83 GB
MTP acceptance rate
—
TTFB
0.54 GB
Power draw
133.5 W

Configuration

Member-level fields are taken from the most recent run

Model
mammoth-1
Quantization
Q5_K_M
Framework
llama.cpp
Version
v0.39
Context length
4096 tokens
Batch size
256
GPU layers
33
Flash Attention
Off

Hardware

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

GPU
Apple M2 Ultra
Measured VRAM (avg)
54.83 GB
OS
macOS 26.1
Power draw
133.5 W

Get started

Original model resources

Base model resources may not match this quantization. Use the verified deployment weights above when available.

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-22

    v0.39 · macOS 26.1 · 4096 ctx

    67.37 tok/s

    Decode

    314.26 tok/s

    Prefill

    2.02 s

    TTFT

    54.83 GB

    VRAM

    —

    MTP

    133.5 W

    Power draw

    L2 batch seed 2026-10-22: llama.cpp on hw_id=103 running model_id=94 at Q5_K_M