Back to search results
gemma-4-26B-A4B-it · Q3_K_M
NVIDIA DGX Spark128GBOllama v0.35L2 Cross-framework verified
This page aggregates 1 real-world runs of gemma-4-26B-A4B-it (Q3_K_M) on NVIDIA DGX Spark with Ollama, contributed by 1 independent source platforms; metrics are averages of published measurements.
113.11 tok/s
Decode
Decode speed
567.16 tok/s
Prefill
Prefill speed
1.59 s
TTFT
Time to first token
39.23 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
5 days ago
Last verified
Performance
- llama.cpp · IQ3_S Decode 112.41 · Prefill 545.82 ·
- llama.cpp · Q3_K_L Decode 105.58 · Prefill 348.6 ·
- Ollama · Q3_K_L Decode 104.18 · Prefill 516.57 ·
- Ollama · Q3_K_M (current)Decode 113.11 · Prefill 567.16 ·
Core figures
- Decode (avg)
- 113.11 tok/s
- Prefill (avg)
- 567.16 tok/s
- TTFT (avg)
- 1.59 s
- VRAM (avg)
- 39.23 GB
- MTP acceptance rate
- —
- TTFB
- 0.9 GB
- Power draw
- 232.6 W
Configuration
Member-level fields are taken from the most recent run
- Model
- gemma-4-26B-A4B-it
- Quantization
- Q3_K_M
- Framework
- Ollama
- Version
- v0.35
- Context length
- 32768 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 DGX Spark
- Nominal VRAM
- 128 GB
- Measured VRAM (avg)
- 39.23 GB
- OS
- Ubuntu 24.04
- Driver
- 565.57
- CUDA
- 12.8
- Power draw
- 232.6 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.35 · Ubuntu 24.04 · CUDA 12.8 · 32768 ctx
113.11 tok/s
Decode
567.16 tok/s
Prefill
1.59 s
TTFT
39.23 GB
VRAM
—
MTP
232.6 W
Power draw
L2 batch seed 2026-10-13: Ollama on hw_id=54 running model_id=64 at Q3_K_M