Qwen-Image-2.1 · NVFP4 rank-128
This page aggregates 1 real-world runs of Qwen-Image-2.1 (NVFP4 rank-128) on NVIDIA RTX 5090 with Nunchaku, contributed by 1 independent source platforms; metrics are averages of published measurements.
4.76 s
Gen time
Per image / per clip
18.74 GB
VRAM
VRAM usage
1
Measured runs
1
Independent sources
Other
Source platforms
6 days ago
Last verified
Performance
No published records for this metric under the same model + hardware yet
Core figures
- Gen time (avg)
- 4.76 s
- VRAM (avg)
- 18.74 GB
- Power draw
- — W
- Output
- 1024x1024 · 25 steps
Configuration
Member-level fields are taken from the most recent run
- Model
- Qwen-Image-2.1
- Quantization
- NVFP4 rank-128
- Framework
- Nunchaku
- Version
- custom runtime (2026-09)
- Batch size
- 1
- Output spec
- 1024x1024 · 25 steps
Hardware
Nominal and measured figures are shown side by side; whether it runs is the reader's call
- GPU
- NVIDIA RTX 5090
- Nominal VRAM
- 32 GB
- Measured VRAM (avg)
- 18.74 GB
- OS
- 未知
Get started
Weights for this deployment
Original model resources
Base model resources may not match this quantization. Use the verified deployment weights above when available.
Deployment guide
https://huggingface.co/mesmertech/Mesmer-Image-21-NunchakuSources & evidence
1 measured records in total, each traceable to its original source
custom runtime (2026-09) · 未知
4.76 s
Gen time
18.74 GB
VRAM
— W
Power draw
RTX 5090，FP4 rank128 transformer + NF4 text/vision encoder + BF16 VAE。1024²、25步、CFG1、eager inference，8次生成耗时中位数4.76s；最高采样板载内存18.74GiB。排除预热、PNG编码、上传、排队和网络；非端到端服务耗时。