Arao AI
Back to search results

Qwen3.5-27B · Q2_K

Start a discussion
Apple M5 Max128GBllama.cpp v0.47L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.5-27B (Q2_K) on Apple M5 Max with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

65.62 tok/s

Decode

Decode speed

279.34 tok/s

Prefill

Prefill speed

0.56 s

TTFT

Time to first token

33.29 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 · Q2_K (current)Decode 65.62 · Prefill 279.34 ·
  2. Ollama · IQ2_M Decode 64.33 · Prefill 209.5 ·
  3. Ollama · IQ3_XS Decode 62.24 · Prefill 282.1 ·
  4. llama.cpp · IQ3_XS Decode 171.72 · Prefill 594.08 ·
  5. llama.cpp · Q4_K_S Decode 146.98 · Prefill 610.13 ·
  6. Ollama · Q5_K_M Decode 119.42 · Prefill 542.35 ·
  7. Ollama · Q6_K Decode 100.18 · Prefill 374.17 ·

Core figures

Decode (avg)
65.62 tok/s
Prefill (avg)
279.34 tok/s
TTFT (avg)
0.56 s
VRAM (avg)
33.29 GB
MTP acceptance rate
—
TTFB
0.52 GB
Power draw
61.3 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.5-27B
Quantization
Q2_K
Framework
llama.cpp
Version
v0.47
Context length
4096 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)
33.29 GB
OS
macOS 26.1
Power draw
61.3 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

    v0.47 · macOS 26.1 · 4096 ctx

    65.62 tok/s

    Decode

    279.34 tok/s

    Prefill

    0.56 s

    TTFT

    33.29 GB

    VRAM

    —

    MTP

    61.3 W

    Power draw

    L2 batch seed 2026-10-19: llama.cpp on hw_id=28 running model_id=50 at Q2_K