Arao AI
Back to search results

Qwen3-4B-Thinking-2507 · IQ2_XS

Start a discussion
Apple M1vLLM v0.39L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3-4B-Thinking-2507 (IQ2_XS) on Apple M1 with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

79.87 tok/s

Decode

Decode speed

436.11 tok/s

Prefill

Prefill speed

1.52 s

TTFT

Time to first token

6.13 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

2 days ago

Last verified

Performance

  1. llama.cpp · IQ2_XS Decode 75.23 · Prefill 381.59 ·
  2. llama.cpp · Q4_K_M Decode 59.36 · Prefill 212.58 ·
  3. vLLM · FP16 Decode 30.09 · Prefill 100.41 ·
  4. vLLM · IQ2_XS (current)Decode 79.87 · Prefill 436.11 ·

Core figures

Decode (avg)
79.87 tok/s
Prefill (avg)
436.11 tok/s
TTFT (avg)
1.52 s
VRAM (avg)
6.13 GB
MTP acceptance rate
—
TTFB
0.64 GB
Power draw
36.2 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3-4B-Thinking-2507
Quantization
IQ2_XS
Framework
vLLM
Version
v0.39
Context length
32768 tokens
Batch size
8
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 M1
Measured VRAM (avg)
6.13 GB
OS
macOS 15.0
Driver
Apple M1
CUDA
N/A
Power draw
36.2 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-16

    v0.39 · macOS 15.0 · CUDA N/A · 32768 ctx

    79.87 tok/s

    Decode

    436.11 tok/s

    Prefill

    1.52 s

    TTFT

    6.13 GB

    VRAM

    —

    MTP

    36.2 W

    Power draw

    L2 batch seed 2026-10-16: vLLM on hw_id=99 running model_id=57 at IQ2_XS