Arao AI
Back to search results

Qwen3.5-0.8B · Q3_K_M

Start a discussion
Apple M5 Pro64GBvLLM v0.44L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.5-0.8B (Q3_K_M) on Apple M5 Pro with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

358.52 tok/s

Decode

Decode speed

1,906.72 tok/s

Prefill

Prefill speed

0.77 s

TTFT

Time to first token

3.24 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 Decode 341.6 · Prefill 1406.03 ·
  2. vLLM · IQ2_M Decode 323.11 · Prefill 1474.14 ·
  3. vLLM · Q3_K_M (current)Decode 358.52 · Prefill 1906.72 ·

Core figures

Decode (avg)
358.52 tok/s
Prefill (avg)
1,906.72 tok/s
TTFT (avg)
0.77 s
VRAM (avg)
3.24 GB
MTP acceptance rate
—
TTFB
2.87 GB
Power draw
116.5 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.5-0.8B
Quantization
Q3_K_M
Framework
vLLM
Version
v0.44
Context length
16384 tokens
Batch size
16
GPU layers
999
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)
3.24 GB
OS
macOS 15.0
Driver
Apple M5 Pro
CUDA
N/A
Power draw
116.5 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

    v0.44 · macOS 15.0 · CUDA N/A · 16384 ctx

    358.52 tok/s

    Decode

    1,906.72 tok/s

    Prefill

    0.77 s

    TTFT

    3.24 GB

    VRAM

    —

    MTP

    116.5 W

    Power draw

    L2 batch seed 2026-10-13: vLLM on hw_id=27 running model_id=51 at Q3_K_M