Back to search results
Qwen3.5-4B · IQ1_S
NVIDIA DGX Spark128GBvLLM v0.34L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3.5-4B (IQ1_S) on NVIDIA DGX Spark with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.
306.37 tok/s
Decode
Decode speed
1,243.41 tok/s
Prefill
Prefill speed
1.79 s
TTFT
Time to first token
5.63 GB
VRAM
VRAM usage
L2 Cross-framework verified
1
Measured runs
1
Independent sources
Other
Source platforms
2 days ago
Last verified
Performance
- vLLM · IQ1_S (current)Decode 306.37 · Prefill 1243.41 ·
- Ollama · IQ2_XS Decode 258.67 · Prefill 1172.95 ·
- vLLM · Q2_K Decode 266.26 · Prefill 1108.22 ·
- vLLM · Q3_K_L Decode 219.42 · Prefill 756.56 ·
Core figures
- Decode (avg)
- 306.37 tok/s
- Prefill (avg)
- 1,243.41 tok/s
- TTFT (avg)
- 1.79 s
- VRAM (avg)
- 5.63 GB
- MTP acceptance rate
- —
- TTFB
- 2.45 GB
- Power draw
- 156.4 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.5-4B
- Quantization
- IQ1_S
- Framework
- vLLM
- Version
- v0.34
- Context length
- 32768 tokens
- Batch size
- 16
- GPU layers
- 999
- 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)
- 5.63 GB
- OS
- Windows 11 23H2
- Driver
- 565.57
- CUDA
- 12.6
- Power draw
- 156.4 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.34 · Windows 11 23H2 · CUDA 12.6 · 32768 ctx
306.37 tok/s
Decode
1,243.41 tok/s
Prefill
1.79 s
TTFT
5.63 GB
VRAM
—
MTP
156.4 W
Power draw
L2 batch seed 2026-10-16: vLLM on hw_id=54 running model_id=48 at IQ1_S