Back to search results
Qwen3-4B · Q3_K_S
NVIDIA DGX Spark128GBStrata v0.49L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3-4B (Q3_K_S) on NVIDIA DGX Spark with Strata, contributed by 1 independent source platforms; metrics are averages of published measurements.
338.04 tok/s
Decode
Decode speed
1,544.38 tok/s
Prefill
Prefill speed
2.43 s
TTFT
Time to first token
7.56 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
2 days ago
Last verified
Performance
- llama.cpp · Q3_K_M Decode 206.5 · Prefill 1076.24 ·
- Strata · Q3_K_S (current)Decode 338.04 · Prefill 1544.38 ·
- Strata · Q5_K_M Decode 221.01 · Prefill 864.66 ·
- vLLM · Q3_K_L Decode 202.71 · Prefill 852.86 ·
- Strata · IQ1_S Decode 323.83 · Prefill 1439.4 ·
Core figures
- Decode (avg)
- 338.04 tok/s
- Prefill (avg)
- 1,544.38 tok/s
- TTFT (avg)
- 2.43 s
- VRAM (avg)
- 7.56 GB
- MTP acceptance rate
- 0.66
- TTFB
- 2.7 GB
- Power draw
- 170.8 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3-4B
- Quantization
- Q3_K_S
- Framework
- Strata
- Version
- v0.49
- Context length
- 16384 tokens
- Batch size
- 256
- GPU layers
- 12
- 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)
- 7.56 GB
- OS
- macOS 15.0
- Driver
- 555.85
- CUDA
- 12.6
- Power draw
- 170.8 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.49 · macOS 15.0 · CUDA 12.6 · 16384 ctx
338.04 tok/s
Decode
1,544.38 tok/s
Prefill
2.43 s
TTFT
7.56 GB
VRAM
0.66
MTP
170.8 W
Power draw
L2 batch seed 2026-10-16: Strata on hw_id=54 running model_id=54 at Q3_K_S