Arao AI
Back to search results

Qwen3.5-27B · Q4_K_M

Start a discussion
NVIDIA RTX PRO 6000 Blackwell 96GB96GBllama.cpp v0.34L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.5-27B (Q4_K_M) on NVIDIA RTX PRO 6000 Blackwell 96GB with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

181.73 tok/s

Decode

Decode speed

687.5 tok/s

Prefill

Prefill speed

0.67 s

TTFT

Time to first token

48.59 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

17 days ago

Last verified

Performance

  1. llama.cpp · Q4_K_M (current)Decode 181.73 · Prefill 687.5 ·
  2. llama.cpp · Q6_K Decode 133.14 · Prefill 542.78 ·
  3. Ollama · FP16 Decode 79.58 · Prefill 274.78 ·
  4. Ollama · Q3_K_L Decode 177.2 · Prefill 930.23 ·

Core figures

Decode (avg)
181.73 tok/s
Prefill (avg)
687.5 tok/s
TTFT (avg)
0.67 s
VRAM (avg)
48.59 GB
MTP acceptance rate
—
TTFB
1.45 GB
Power draw
359.9 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.5-27B
Quantization
Q4_K_M
Framework
llama.cpp
Version
v0.34
Context length
4096 tokens
Batch size
512
GPU layers
33
Flash Attention
On

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)
48.59 GB
OS
Windows 11 23H2
Driver
560.94
CUDA
12.4
Power draw
359.9 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

    v0.34 · Windows 11 23H2 · CUDA 12.4 · 4096 ctx

    181.73 tok/s

    Decode

    687.5 tok/s

    Prefill

    0.67 s

    TTFT

    48.59 GB

    VRAM

    —

    MTP

    359.9 W

    Power draw

    L2 batch seed 2026-10-19: llama.cpp on hw_id=68 running model_id=50 at Q4_K_M