Arao AI
Back to search results

DeepSeek-R1-Distill-Qwen-14B · IQ2_M

Start a discussion
NVIDIA DGX Spark128GBOllama v0.51L2 Cross-framework verified

This page aggregates 1 real-world runs of DeepSeek-R1-Distill-Qwen-14B (IQ2_M) on NVIDIA DGX Spark with Ollama, contributed by 1 independent source platforms; metrics are averages of published measurements.

126.86 tok/s

Decode

Decode speed

519.13 tok/s

Prefill

Prefill speed

0.36 s

TTFT

Time to first token

17.39 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

5 days ago

Last verified

Performance

  1. vLLM · Q3_K_L Decode 126.48 · Prefill 583.23 ·
  2. vLLM · Q4_K_M Decode 106.85 · Prefill 369.96 ·
  3. Ollama · IQ2_M (current)Decode 126.86 · Prefill 519.13 ·

Core figures

Decode (avg)
126.86 tok/s
Prefill (avg)
519.13 tok/s
TTFT (avg)
0.36 s
VRAM (avg)
17.39 GB
MTP acceptance rate
—
TTFB
1.01 GB
Power draw
278.5 W

Configuration

Member-level fields are taken from the most recent run

Model
DeepSeek-R1-Distill-Qwen-14B
Quantization
IQ2_M
Framework
Ollama
Version
v0.51
Context length
4096 tokens
Batch size
256
GPU layers
20
Flash Attention
On

Hardware

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

GPU
NVIDIA DGX Spark
Nominal VRAM
128 GB
Measured VRAM (avg)
17.39 GB
OS
Ubuntu 24.04
Driver
555.85
CUDA
12.8
Power draw
278.5 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-13

    v0.51 · Ubuntu 24.04 · CUDA 12.8 · 4096 ctx

    126.86 tok/s

    Decode

    519.13 tok/s

    Prefill

    0.36 s

    TTFT

    17.39 GB

    VRAM

    —

    MTP

    278.5 W

    Power draw

    L2 batch seed 2026-10-13: Ollama on hw_id=54 running model_id=71 at IQ2_M