Arao AI
Back to search results

Qwen3-4B-Thinking-2507 · IQ3_S

Start a discussion
NVIDIA RTX 308010GBStrata v0.20L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3-4B-Thinking-2507 (IQ3_S) on NVIDIA RTX 3080 with Strata, contributed by 1 independent source platforms; metrics are averages of published measurements.

134.76 tok/s

Decode

Decode speed

504.77 tok/s

Prefill

Prefill speed

0.28 s

TTFT

Time to first token

7.64 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

5 days ago

Last verified

Performance

  1. Ollama · Q2_K Decode 93.59 · Prefill 336.53 ·
  2. Ollama · Q4_K_S Decode 73.95 · Prefill 380.29 ·
  3. Strata · IQ3_S (current)Decode 134.76 · Prefill 504.77 ·

Core figures

Decode (avg)
134.76 tok/s
Prefill (avg)
504.77 tok/s
TTFT (avg)
0.28 s
VRAM (avg)
7.64 GB
MTP acceptance rate
0.51
TTFB
1.08 GB
Power draw
365.8 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3-4B-Thinking-2507
Quantization
IQ3_S
Framework
Strata
Version
v0.20
Context length
32768 tokens
Batch size
1024
GPU layers
15
Flash Attention
On

Hardware

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

GPU
NVIDIA RTX 3080
Nominal VRAM
10 GB
Measured VRAM (avg)
7.64 GB
OS
Ubuntu 24.04
Driver
565.57
CUDA
12.6
Power draw
365.8 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

    v0.20 · Ubuntu 24.04 · CUDA 12.6 · 32768 ctx

    134.76 tok/s

    Decode

    504.77 tok/s

    Prefill

    0.28 s

    TTFT

    7.64 GB

    VRAM

    0.51

    MTP

    365.8 W

    Power draw

    L2 batch seed 2026-10-13: Strata on hw_id=44 running model_id=57 at IQ3_S