Arao AI
Back to search results

Qwen3-4B · FP16

Start a discussion
Qualcomm Snapdragon X EliteStrata v0.26L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3-4B (FP16) on Qualcomm Snapdragon X Elite with Strata, contributed by 1 independent source platforms; metrics are averages of published measurements.

39.4 tok/s

Decode

Decode speed

159.35 tok/s

Prefill

Prefill speed

2.29 s

TTFT

Time to first token

15.2 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 · IQ4_XS Decode 66.12 · Prefill 360.69 ·
  2. vLLM · Q3_K_M Decode 74.16 · Prefill 385.13 ·
  3. Strata · FP16 (current)Decode 39.4 · Prefill 159.35 ·

Core figures

Decode (avg)
39.4 tok/s
Prefill (avg)
159.35 tok/s
TTFT (avg)
2.29 s
VRAM (avg)
15.2 GB
MTP acceptance rate
0.66
TTFB
0.32 GB
Power draw
34 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3-4B
Quantization
FP16
Framework
Strata
Version
v0.26
Context length
16384 tokens
Batch size
1024
GPU layers
12
Flash Attention
Off

Hardware

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

GPU
Qualcomm Snapdragon X Elite
Measured VRAM (avg)
15.2 GB
OS
Windows 11 23H2
Driver
Qualcomm Adreno 7.0
CUDA
N/A
Power draw
34 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

    v0.26 · Windows 11 23H2 · CUDA N/A · 16384 ctx

    39.4 tok/s

    Decode

    159.35 tok/s

    Prefill

    2.29 s

    TTFT

    15.2 GB

    VRAM

    0.66

    MTP

    34 W

    Power draw

    L2 batch seed 2026-10-16: Strata on hw_id=104 running model_id=54 at FP16