Qwen-Image-2.1 vs sdxl-turbo
Qwen-Image-2.1 (7.1B parameters) and sdxl-turbo (2.6B parameters) side by side: the memory each needs at every precision, what it costs to run on a live GPU, and the context window, KV cache and license where they are published. Numbers are computed from the models' published specs; this page does not rank quality.
Side by side
| Fact | Qwen-Image-2.1 | sdxl-turbo |
|---|---|---|
| Parameters | 7.1B | 2.6B |
| Architecture | – | – |
| License | – | – |
| Published precision | BF16 | F32 |
| VRAM needed, As published | 15.9 GB | 11.5 GB |
| VRAM needed, FP8 | 8.0 GB | 2.9 GB |
| VRAM needed, INT4 | 4.0 GB | 1.4 GB |
| Cheapest live fit, As published | RTX A4000 · $0.167/hr | V100 · $0.088/hr |
| Cheapest live fit, FP8 | RTX 4070 Super · $0.121/hr | RTX 4070 Super · $0.121/hr |
| Cheapest live fit, INT4 | RTX 4070 Super · $0.121/hr | RTX 4070 Super · $0.121/hr |
VRAM is the weight size at each precision times a flat 1.2 overhead; see the methodology. The FP8 and INT4 rows need a quantized checkpoint or an engine that quantizes on load. The fit is the lowest-priced single GPU type that holds the model at that precision, or the lowest-priced multi-GPU set (up to 8) when none does.
Which to pick
- sdxl-turbo needs less VRAM at its published precision (11.5 GB against 15.9 GB), so it fits on a smaller GPU.
- sdxl-turbo has the cheaper live GPU fit at its published precision ($0.088/hr against $0.167/hr).
These follow only from the facts in the table above. Whether either model does your task well is a separate question this page does not answer.
Keep reading
- Qwen-Image-2.1: full VRAM table and live GPU fit
- sdxl-turbo: full VRAM table and live GPU fit
- The Qwen model series
- The Stable Diffusion model series
- All models that fit in 16 GB
- All models that fit in 12 GB
Other comparisons
