Arao AI
Back to search results

Qwen3-Embedding-0.6B · IQ3_S

Start a discussion
AMD Strix Halo (Radeon 8060S)vLLM v0.38L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3-Embedding-0.6B (IQ3_S) on AMD Strix Halo (Radeon 8060S) with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

267.01 tok/s

Decode

Decode speed

1,239.81 tok/s

Prefill

Prefill speed

0.37 s

TTFT

Time to first token

2.93 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

2 days ago

Last verified

Performance

  1. vLLM · IQ2_XS Decode 355.91 · Prefill 1759.4 ·
  2. vLLM · IQ3_S (current)Decode 267.01 · Prefill 1239.81 ·
  3. Strata · IQ3_S Decode 374.22 · Prefill 1937.23 ·

Core figures

Decode (avg)
267.01 tok/s
Prefill (avg)
1,239.81 tok/s
TTFT (avg)
0.37 s
VRAM (avg)
2.93 GB
MTP acceptance rate
—
TTFB
2.14 GB
Power draw
97.7 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3-Embedding-0.6B
Quantization
IQ3_S
Framework
vLLM
Version
v0.38
Context length
16384 tokens
Batch size
8
GPU layers
999
Flash Attention
Off

Hardware

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

GPU
AMD Strix Halo (Radeon 8060S)
Measured VRAM (avg)
2.93 GB
OS
Windows 11 23H2
Driver
amdgpu-pro 24.20
CUDA
N/A
Power draw
97.7 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

    v0.38 · Windows 11 23H2 · CUDA N/A · 16384 ctx

    267.01 tok/s

    Decode

    1,239.81 tok/s

    Prefill

    0.37 s

    TTFT

    2.93 GB

    VRAM

    —

    MTP

    97.7 W

    Power draw

    L2 batch seed 2026-10-16: vLLM on hw_id=87 running model_id=62 at IQ3_S