Arao AI
Back to search results

Qwen3.6-27B · FP16

Start a discussion
NVIDIA RTX Spark128GBllama.cpp v0.41L2 Cross-framework verified

This page aggregates 1 real-world runs of Qwen3.6-27B (FP16) on NVIDIA RTX Spark with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

5.5 tok/s

Decode

Decode speed

19.76 tok/s

Prefill

Prefill speed

1.91 s

TTFT

Time to first token

107 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 · FP16 (current)Decode 5.5 · Prefill 19.76 ·
  2. llama.cpp · Q5_K_M Decode 8.41 · Prefill 32.22 ·
  3. llama.cpp · Q8_0 Decode 7.68 · Prefill 32.82 ·
  4. Ollama · Q5_K_M Decode 8.83 · Prefill 30.38 ·
  5. Ollama · Q8_0 Decode 6.72 · Prefill 25.96 ·
  6. llama.cpp · IQ3_S Decode 13.81 · Prefill 45.91 ·
  7. llama.cpp · Q4_K_M Decode 10.95 · Prefill 50.01 ·
  8. Ollama · Q4_K_M Decode 10.39 · Prefill 39.74 ·

Core figures

Decode (avg)
5.5 tok/s
Prefill (avg)
19.76 tok/s
TTFT (avg)
1.91 s
VRAM (avg)
107 GB
MTP acceptance rate
—
TTFB
0.04 GB
Power draw
50.5 W

Configuration

Member-level fields are taken from the most recent run

Model
Qwen3.6-27B
Quantization
FP16
Framework
llama.cpp
Version
v0.41
Context length
16384 tokens
Batch size
256
GPU layers
33
Flash Attention
Off

Hardware

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

GPU
NVIDIA RTX Spark
Nominal VRAM
128 GB
Measured VRAM (avg)
107 GB
OS
Ubuntu 24.04
Driver
615.78
CUDA
13.0
Power draw
50.5 W
Reference price
$2,399 · updated 2026-10-19

Prices are not live; check the store page

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 2026-10-20

    v0.41 · Ubuntu 24.04 · CUDA 13.0 · 16384 ctx

    5.5 tok/s

    Decode

    19.76 tok/s

    Prefill

    1.91 s

    TTFT

    107 GB

    VRAM

    —

    MTP

    50.5 W

    Power draw

    L2 batch seed 2026-10-20 (Linux 116GiB unlock): llama.cpp on Spark running model_id=6 at FP16