[setsewerd]

Since the ranking is based on token usage, wouldn't this ranking be skewed by the fact that small models' APIs are often used for consumer products, especially free ones? Meanwhile reasoning models skew it in the opposite direction, but to what extent I don't know.

It's an interesting proxy, but idk how reliable it'd be.

[matznerd]

Also, these small models are meant to be run local so not going to appear on openrouter...