Arao AI
Back to search results

mammoth-1 · IQ2_M

Start a discussion
Apple M2 Ultrallama.cpp v0.38L2 Cross-framework verified

This page aggregates 1 real-world runs of mammoth-1 (IQ2_M) on Apple M2 Ultra with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

102.29 tok/s

Decode

Decode speed

417.96 tok/s

Prefill

Prefill speed

1.66 s

TTFT

Time to first token

31.58 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

14 days ago

Last verified

Performance

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

Core figures

Decode (avg)
102.29 tok/s
Prefill (avg)
417.96 tok/s
TTFT (avg)
1.66 s
VRAM (avg)
31.58 GB
MTP acceptance rate
—
TTFB
0.82 GB
Power draw
129.6 W

Configuration

Member-level fields are taken from the most recent run

Model
mammoth-1
Quantization
IQ2_M
Framework
llama.cpp
Version
v0.38
Context length
32768 tokens
Batch size
512
GPU layers
28
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)
31.58 GB
OS
macOS 26.1
Power draw
129.6 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

    v0.38 · macOS 26.1 · 32768 ctx

    102.29 tok/s

    Decode

    417.96 tok/s

    Prefill

    1.66 s

    TTFT

    31.58 GB

    VRAM

    —

    MTP

    129.6 W

    Power draw

    L2 batch seed 2026-10-22: llama.cpp on hw_id=103 running model_id=94 at IQ2_M