Arao AI
Back to search results

Llama-3.1-8B-Instruct · IQ3_XS

Start a discussion
Apple M5 Max128GBOllama v0.41L2 Cross-framework verified

This page aggregates 1 real-world runs of Llama-3.1-8B-Instruct (IQ3_XS) on Apple M5 Max with Ollama, contributed by 1 independent source platforms; metrics are averages of published measurements.

82.82 tok/s

Decode

Decode speed

392.75 tok/s

Prefill

Prefill speed

0.11 s

TTFT

Time to first token

12.52 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

16 days ago

Last verified

Performance

  1. llama.cpp · IQ2_M Decode 101.46 · Prefill 460.44 ·
  2. llama.cpp · Q3_K_L Decode 81.19 · Prefill 358.64 ·
  3. Ollama · IQ3_XS (current)Decode 82.82 · Prefill 392.75 ·
  4. Ollama · Q3_K_S Decode 79.81 · Prefill 418 ·

Core figures

Decode (avg)
82.82 tok/s
Prefill (avg)
392.75 tok/s
TTFT (avg)
0.11 s
VRAM (avg)
12.52 GB
MTP acceptance rate
—
TTFB
0.66 GB
Power draw
54.2 W

Configuration

Member-level fields are taken from the most recent run

Model
Llama-3.1-8B-Instruct
Quantization
IQ3_XS
Framework
Ollama
Version
v0.41
Context length
16384 tokens
Batch size
256
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 M5 Max
Nominal VRAM
128 GB
Measured VRAM (avg)
12.52 GB
OS
macOS 26.1
Power draw
54.2 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-19

    v0.41 · macOS 26.1 · 16384 ctx

    82.82 tok/s

    Decode

    392.75 tok/s

    Prefill

    0.11 s

    TTFT

    12.52 GB

    VRAM

    —

    MTP

    54.2 W

    Power draw

    L2 batch seed 2026-10-19: Ollama on hw_id=28 running model_id=68 at IQ3_XS