Arao AI
Back to search results

Qwen3-4B · Q2_K

Start a discussion
Apple M1Ollama v0.22L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3-4B (Q2_K) on Apple M1 with Ollama, contributed by 1 independent source platforms; metrics are averages of published measurements.

72.5 tok/s

Decode

Decode speed

393.97 tok/s

Prefill

Prefill speed

1.2 s

TTFT

Time to first token

6.72 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 · Q3_K_L Decode 65.56 · Prefill 344.27 ·
  2. llama.cpp · Q8_0 Decode 36.81 · Prefill 167.78 ·
  3. Ollama · Q2_K (current)Decode 72.5 · Prefill 393.97 ·

Core figures

Decode (avg)
72.5 tok/s
Prefill (avg)
393.97 tok/s
TTFT (avg)
1.2 s
VRAM (avg)
6.72 GB
MTP acceptance rate
—
TTFB
0.58 GB
Power draw
26 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3-4B
Quantization
Q2_K
Framework
Ollama
Version
v0.22
Context length
2048 tokens
Batch size
256
GPU layers
99
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)
6.72 GB
OS
macOS 15.0
Driver
Apple M1
CUDA
N/A
Power draw
26 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

    v0.22 · macOS 15.0 · CUDA N/A · 2048 ctx

    72.5 tok/s

    Decode

    393.97 tok/s

    Prefill

    1.2 s

    TTFT

    6.72 GB

    VRAM

    —

    MTP

    26 W

    Power draw

    L2 batch seed 2026-10-16: Ollama on hw_id=99 running model_id=54 at Q2_K