Arao AI
Back to search results

Qwen3.5-0.8B · IQ1_S

Start a discussion
Apple M5 Pro64GBllama.cpp v0.37L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.5-0.8B (IQ1_S) on Apple M5 Pro with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

341.6 tok/s

Decode

Decode speed

1,406.03 tok/s

Prefill

Prefill speed

2.42 s

TTFT

Time to first token

2.8 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

5 days ago

Last verified

Performance

  1. llama.cpp · IQ1_S (current)Decode 341.6 · Prefill 1406.03 ·
  2. vLLM · IQ2_M Decode 323.11 · Prefill 1474.14 ·
  3. vLLM · Q3_K_M Decode 358.52 · Prefill 1906.72 ·

Core figures

Decode (avg)
341.6 tok/s
Prefill (avg)
1,406.03 tok/s
TTFT (avg)
2.42 s
VRAM (avg)
2.8 GB
MTP acceptance rate
—
TTFB
2.73 GB
Power draw
55 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.5-0.8B
Quantization
IQ1_S
Framework
llama.cpp
Version
v0.37
Context length
32768 tokens
Batch size
512
GPU layers
20
Flash Attention
On

Hardware

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

GPU
Apple M5 Pro
Nominal VRAM
64 GB
Measured VRAM (avg)
2.8 GB
OS
macOS 15.0
Driver
Apple M5 Pro
CUDA
N/A
Power draw
55 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-13

    v0.37 · macOS 15.0 · CUDA N/A · 32768 ctx

    341.6 tok/s

    Decode

    1,406.03 tok/s

    Prefill

    2.42 s

    TTFT

    2.8 GB

    VRAM

    —

    MTP

    55 W

    Power draw

    L2 batch seed 2026-10-13: llama.cpp on hw_id=27 running model_id=51 at IQ1_S