Back to search results
Qwen3.5-4B · Q3_K_S
NVIDIA H100 80GB HBM380GBvLLM v0.34L2 Cross-framework verified
This page aggregates 1 real-world runs of Qwen3.5-4B (Q3_K_S) on NVIDIA H100 80GB HBM3 with vLLM, contributed by 1 independent source platforms; metrics are averages of published measurements.
362.27 tok/s
Decode
Decode speed
1,696.48 tok/s
Prefill
Prefill speed
1.81 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
16 days ago
Last verified
Performance
- vLLM · IQ3_S Decode 329.61 · Prefill 1536.04 ·
- vLLM · Q3_K_S (current)Decode 362.27 · Prefill 1696.48 ·
- Strata · Q2_K Decode 373.13 · Prefill 1907.75 ·
Core figures
- Decode (avg)
- 362.27 tok/s
- Prefill (avg)
- 1,696.48 tok/s
- TTFT (avg)
- 1.81 s
- VRAM (avg)
- 7.56 GB
- MTP acceptance rate
- —
- TTFB
- 2.9 GB
- Power draw
- 571.5 W
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen3.5-4B
- Quantization
- Q3_K_S
- Framework
- vLLM
- Version
- v0.34
- Context length
- 4096 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 H100 80GB HBM3
- Nominal VRAM
- 80 GB
- Measured VRAM (avg)
- 7.56 GB
- OS
- Ubuntu 22.04
- Driver
- 560.94
- CUDA
- 12.8
- Power draw
- 571.5 W
Sources & evidence
1 measured records in total, each traceable to its original source
v0.34 · Ubuntu 22.04 · CUDA 12.8 · 4096 ctx
362.27 tok/s
Decode
1,696.48 tok/s
Prefill
1.81 s
TTFT
7.56 GB
VRAM
—
MTP
571.5 W
Power draw
L2 batch seed 2026-10-19: vLLM on hw_id=55 running model_id=48 at Q3_K_S