Arao AI
Back to search results

Qwen3.6-27B · IQ3_S

Start a discussion
NVIDIA RTX PRO 6000 Blackwell 96GB96GBvLLM v0.45L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.6-27B (IQ3_S) on NVIDIA RTX PRO 6000 Blackwell 96GB with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.

220.08 tok/s

Decode

Decode speed

989.75 tok/s

Prefill

Prefill speed

0.29 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

16 days ago

Last verified

Performance

  1. llama.cpp · IQ1_S Decode 267.34 · Prefill 934.02 ·
  2. llama.cpp · Q8_0 Decode 105.75 · Prefill 445.38 ·
  3. vLLM · IQ3_S (current)Decode 220.08 · Prefill 989.75 ·

Core figures

Decode (avg)
220.08 tok/s
Prefill (avg)
989.75 tok/s
TTFT (avg)
0.29 s
VRAM (avg)
39.52 GB
MTP acceptance rate
—
TTFB
1.76 GB
Power draw
318.1 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.6-27B
Quantization
IQ3_S
Framework
vLLM
Version
v0.45
Context length
8192 tokens
Batch size
32
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
NVIDIA RTX PRO 6000 Blackwell 96GB
Nominal VRAM
96 GB
Measured VRAM (avg)
39.52 GB
OS
Ubuntu 22.04
Driver
572.83
CUDA
12.8
Power draw
318.1 W

Get started

Original model resources

Base model resources may not match this quantization. Use the verified deployment weights above when available.

Sources & evidence

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

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

    v0.45 · Ubuntu 22.04 · CUDA 12.8 · 8192 ctx

    220.08 tok/s

    Decode

    989.75 tok/s

    Prefill

    0.29 s

    TTFT

    39.52 GB

    VRAM

    —

    MTP

    318.1 W

    Power draw

    L2 batch seed 2026-10-19: vLLM on hw_id=68 running model_id=6 at IQ3_S