Arao AI
Back to search results

Qwen3.5-27B · IQ3_S

Start a discussion
NVIDIA H2097GBStrata v0.30L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.5-27B (IQ3_S) on NVIDIA H20 with Strata, contributed by 1 independent source platforms; metrics are averages of published measurements.

67.46 tok/s

Decode

Decode speed

225.71 tok/s

Prefill

Prefill speed

0.85 s

TTFT

Time to first token

39.52 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

17 days ago

Last verified

Performance

  1. vLLM · IQ2_XS Decode 63.17 · Prefill 260.61 ·
  2. vLLM · Q8_0 Decode 27.55 · Prefill 97.34 ·
  3. Strata · IQ3_S (current)Decode 67.46 · Prefill 225.71 ·

Core figures

Decode (avg)
67.46 tok/s
Prefill (avg)
225.71 tok/s
TTFT (avg)
0.85 s
VRAM (avg)
39.52 GB
MTP acceptance rate
0.64
TTFB
0.54 GB
Power draw
346.8 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.5-27B
Quantization
IQ3_S
Framework
Strata
Version
v0.30
Context length
4096 tokens
Batch size
1024
GPU layers
5
Flash Attention
On

Hardware

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

GPU
NVIDIA H20
Nominal VRAM
97 GB
Measured VRAM (avg)
39.52 GB
OS
Ubuntu 22.04
Driver
572.83
CUDA
12.4
Power draw
346.8 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-19

    v0.30 · Ubuntu 22.04 · CUDA 12.4 · 4096 ctx

    67.46 tok/s

    Decode

    225.71 tok/s

    Prefill

    0.85 s

    TTFT

    39.52 GB

    VRAM

    0.64

    MTP

    346.8 W

    Power draw

    L2 batch seed 2026-10-19: Strata on hw_id=56 running model_id=50 at IQ3_S