How Are Models Ranked?

The ranking system evaluates models based on several key factors to ensure that the best options are highlighted. The goal is to reflect not just popularity but also efficiency, reliability, and how actively the models are maintained.

Popularity: Models with more likes and downloads score higher, reflecting real-world usage and interest from users and developers.

Recency: Models that receive regular updates rank higher to ensure actively maintained and relevant models are prioritized.

Efficiency: Efficiency: Models using less RAM and processing power are rewarded for being lightweight and easy to integrate.

Benchmarks: Scores from external benchmarks are included, adding an extra layer of performance validation.

To ensure fairness, each metric is adjusted so that a model's score is relative to other models, making comparisons consistent across the board. The final score combines all these factors, with some metrics being more important than others based on their impact on how useful the model is. The data for these calculations, such as likes, downloads, and performance benchmarks, comes from reliable third-party sources to keep the rankings objective, accurate, and up-to-date.