Arao AI
Back to search results

Raon-1-mini-preview · Q4_K_M

Start a discussion
NVIDIA RTX Spark128GBllama.cpp v0.38L2 Cross-framework verified

This page aggregates 2 real-world runs of Raon-1-mini-preview (Q4_K_M) on NVIDIA RTX Spark with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

47.78 tok/s

Decode

Decode speed

218.5 tok/s

Prefill

Prefill speed

2.01 s

TTFT

Time to first token

51.69 GB

VRAM

VRAM usage

L2 Cross-framework verified

2

Measured runs

1

Independent sources

Other

Source platforms

13 days ago

Last verified

Performance

  1. llama.cpp · Q4_K_M (current)Decode 47.78 · Prefill 218.5 ·
  2. Ollama · IQ4_XS Decode 49.65 · Prefill 223.79 ·
  3. Ollama · Q4_K_M Decode 45.05 · Prefill 148.34 ·

Core figures

Decode (avg)
47.78 tok/s
Prefill (avg)
218.5 tok/s
TTFT (avg)
2.01 s
VRAM (avg)
51.69 GB
MTP acceptance rate
—
TTFB
0.38 GB
Power draw
63.55 W

Configuration

Member-level fields are taken from the most recent run

Model
Raon-1-mini-preview
Quantization
Q4_K_M
Framework
llama.cpp
Version
v0.38
Context length
16384 tokens
Batch size
256
GPU layers
28
Flash Attention
On

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)
51.69 GB
OS
Ubuntu 24.04
Driver
615.78
CUDA
13.0
Power draw
54.8 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

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-22

    v0.38 · Ubuntu 24.04 · CUDA 13.0 · 16384 ctx

    49.11 tok/s

    Decode

    252.26 tok/s

    Prefill

    2.35 s

    TTFT

    51.69 GB

    VRAM

    —

    MTP

    54.8 W

    Power draw

    L2 batch seed 2026-10-22: llama.cpp on hw_id=110 running model_id=93 at Q4_K_M
  2. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-22

    v0.24 · Ubuntu 24.04 · CUDA 13.0 · 4096 ctx

    46.44 tok/s

    Decode

    184.73 tok/s

    Prefill

    1.66 s

    TTFT

    51.69 GB

    VRAM

    —

    MTP

    72.3 W

    Power draw

    L2 batch seed 2026-10-22: llama.cpp on hw_id=110 running model_id=93 at Q4_K_M