Arao AI
Back to search results

Hy-Memory-30B-A3B · IQ4_XS

Start a discussion
NVIDIA RTX Spark128GBllama.cpp v0.23L2 Cross-framework verified

This page aggregates 1 real-world runs of Hy-Memory-30B-A3B (IQ4_XS) on NVIDIA RTX Spark with llama.cpp, contributed by 1 independent source platforms; metrics are averages of published measurements.

93.67 tok/s

Decode

Decode speed

407.31 tok/s

Prefill

Prefill speed

2.46 s

TTFT

Time to first token

48.72 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 Decode 42.81 · Prefill 164.71 ·
  2. llama.cpp · Q5_K_M Decode 67.28 · Prefill 348.21 ·
  3. llama.cpp · Q8_0 Decode 58.7 · Prefill 210.79 ·
  4. Ollama · Q5_K_M Decode 64.27 · Prefill 274.24 ·
  5. Ollama · Q8_0 Decode 46.83 · Prefill 233.52 ·
  6. llama.cpp · IQ4_XS (current)Decode 93.67 · Prefill 407.31 ·
  7. llama.cpp · Q4_K_M Decode 85.75 · Prefill 321.23 ·
  8. Ollama · IQ4_XS Decode 76.19 · Prefill 410.3 ·
  9. Ollama · Q4_K_M Decode 69.54 · Prefill 373.27 ·

Core figures

Decode (avg)
93.67 tok/s
Prefill (avg)
407.31 tok/s
TTFT (avg)
2.46 s
VRAM (avg)
48.72 GB
MTP acceptance rate
—
TTFB
0.75 GB
Power draw
75.1 W

Configuration

Member-level fields are taken from the most recent run

Model
Hy-Memory-30B-A3B
Quantization
IQ4_XS
Framework
llama.cpp
Version
v0.23
Context length
2048 tokens
Batch size
512
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)
48.72 GB
OS
Windows 11 24H2
Driver
NVIDIA 590.10
CUDA
13.0
Power draw
75.1 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 2026-10-19

    v0.23 · Windows 11 24H2 · CUDA 13.0 · 2048 ctx

    93.67 tok/s

    Decode

    407.31 tok/s

    Prefill

    2.46 s

    TTFT

    48.72 GB

    VRAM

    —

    MTP

    75.1 W

    Power draw

    L2 batch seed 2026-10-19: llama.cpp on hw_id=110 running model_id=87 at IQ4_XS