About SD Benchmark

SD Benchmark serves as a dedicated platform for analyzing and comparing Stable Diffusion performance across diverse hardware and software environments. By aggregating thousands of community-contributed benchmarks, we empower users—ranging from researchers and developers to AI enthusiasts—to evaluate and optimize their setups effectively. Our intuitive tools deliver actionable insights into real-world performance, helping you select the best configurations for your projects.

Benchmark Data Sources

Our benchmark dataset is sourced from the community-driven SD System Info extension, a trusted hub for global performance data. This collaboration ensures a diverse dataset reflecting real scenarios. Visit the primary source to contribute: https://vladmandic.github.io/sd-extension-system-info/pages/benchmark.html

Understanding Inference Accuracy

We use averaging to create reliable metrics from multiple runs, accounting for variations and enabling fair comparisons. For example, with it/s values of 6, 3, and 8:

This helps assess typical performance and guide decisions.