I like that https://artificialanalysis.ai/leaderboards/models https://artificialanalysis.ai/leaderboards/models describes both quality and speed (tokens/s and f

I like that https://artificialanalysis.ai/leaderboards/models https://artificialanalysis.ai/leaderboards/models describes both quality and speed (tokens/s and first chunk s). Not sure how accurate it is; anyone know? Speed and variance of it in particular seems difficult to pin down because providers obviously vary it with load to control their costs.