Qwen-Image: local inference speed (tok/s), sourced
As of 2026-11-14, this dataset has 13 recipes for Qwen-Image. 7 were measured by the poster. These are image, video, or audio results. Speed is in the metric column, not tok/s. 2 entries have a date, from 2026-10-11 to 2026-10-30.
Filter these in the explorer →
| Model | Engine | Hardware | tok/s | Metric | Source level | Date | Source |
|---|---|---|---|---|---|---|---|
| Qwen-Image 2.1 (7B + 8B text encoder) | ComfyUI | RTX 4090 24GB | 8.1 s / 1024² (40 steps); 30.2 s / 2048×1152 (50 steps); peak 16.2GB | Measured | link | ||
| Qwen-Image 2.1 | TensorFold | RTX 5090 @450W | 4.53 s / 6.94 s / 13.87 s (1024², 25 steps; NVFP4 peak 22.9GB; images not identical to Q6) | Measured | link | ||
| Qwen-Image 2.1 | Unspecified | RTX 5090 32GB (455W power limit) | Qwen-Image 2.1 base model, int8, 25 steps, 1MP: 5.2 s per image (median of the last 200 images), about 687 images per hour; power 399 W (nvidia-smi 60 s average); energy 0.58 Wh per image | Measured | 2026-10-11 | link | |
| Qwen-Image 2.1 (edit) | ComfyUI | Mac M4 Pro | 3 min 52 s / image (768 edit) | Measured | 2026-10-30 | link | |
| Qwen-Image-2.1-Turbo (official 8 steps) | Unspecified | RTX 5090 | 29 s vs 108 s / image (2048²) | Measured | link | ||
| Qwen-Image-2.1-Turbo | Unspecified | RTX 3060 | ~35 s / image (8 steps) | Measured | link | ||
| Z-Image Turbo vs Qwen-Image 2.1 | Unspecified | RTX 5090 | Z-Image ~2× faster (no absolute times); quality: Qwen wins 9:6 | Measured | link | ||
| Qwen-Image 2.1 | Unspecified | RTX 3090 | 82–96 s / 2K (public receipt) | Relayed | link | ||
| Qwen-Image 2.1 | Unspecified | RTX 5090 32GB | 5.93 s; NVFP4 6.7 s vs INT8 7.6 s vs BF16 15.6 s (1024²) | Relayed | link | ||
| Qwen-Image 2.1 | Unspecified | RTX 5090 | 13.5 s → 1.9 s / image (vendor claim) | Claimed | link | ||
| Qwen-Image 2.1 | Unspecified | RTX 4060 8GB / 3060 12GB / 5070 Ti 16GB | ~2 min / ~40 s / ~25 s (1024², 25 steps) | Relayed | link | ||
| Qwen-Image 2.1 | SGLang | 1× DGX Spark (vs 4090) | 52.6 s / 1024²; SGLang table: Spark 35.4 s, 4090 18.7 s | Relayed | link | ||
| Qwen-Image 2.1 | mflux | Mac M5 Max | ~78 s / 1024² (40 steps) | Relayed | link |
Numbers come from each source. This project did not measure them again. Models and settings differ between rows. Source levels: Official = project or vendor; Measured = the poster ran it (an engine or recipe author must also publish raw data and reproduction steps); Relayed = second-hand; Claimed = prediction or no reproduction details.