Arao AI
Back to search results

gemma-4-E2B-it · Q3_K_L

Start a discussion
NVIDIA DGX Spark128GBStrata v0.55L2 Cross-framework verified

This page aggregates 1 real-world runs of gemma-4-E2B-it (Q3_K_L) on NVIDIA DGX Spark with Strata, contributed by 1 independent source platforms; metrics are averages of published measurements.

389.31 tok/s

Decode

Decode speed

1,363.67 tok/s

Prefill

Prefill speed

1.89 s

TTFT

Time to first token

5.12 GB

VRAM

VRAM usage

L2 Cross-framework verified

1

Measured runs

1

Independent sources

Other

Source platforms

2 days ago

Last verified

Performance

  1. vLLM · IQ2_M Decode 380.91 · Prefill 2066.12 ·
  2. vLLM · IQ3_XS Decode 360.01 · Prefill 1436.84 ·
  3. Strata · Q3_K_L (current)Decode 389.31 · Prefill 1363.67 ·
  4. Strata · Q4_K_M Decode 326.71 · Prefill 1358.52 ·

Core figures

Decode (avg)
389.31 tok/s
Prefill (avg)
1,363.67 tok/s
TTFT (avg)
1.89 s
VRAM (avg)
5.12 GB
MTP acceptance rate
0.59
TTFB
3.11 GB
Power draw
253.6 W

Configuration

Member-level fields are taken from the most recent run

Model
gemma-4-E2B-it
Quantization
Q3_K_L
Framework
Strata
Version
v0.55
Context length
2048 tokens
Batch size
1024
GPU layers
8
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)
5.12 GB
OS
Ubuntu 22.04
Driver
555.85
CUDA
12.6
Power draw
253.6 W

Sources & evidence

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

  1. L2 Cross-framework verifiedOtherOriginal link Verified on 2026-10-16

    v0.55 · Ubuntu 22.04 · CUDA 12.6 · 2048 ctx

    389.31 tok/s

    Decode

    1,363.67 tok/s

    Prefill

    1.89 s

    TTFT

    5.12 GB

    VRAM

    0.59

    MTP

    253.6 W

    Power draw

    L2 batch seed 2026-10-16: Strata on hw_id=54 running model_id=66 at Q3_K_L