Re:State of AI: An Empirical 100T Token Study with OpenRouter
by YetAnotherNick on Friday December 05, 2025 @03:00AM
(#46157856)
The bigger issue is that they count small based on fixed number of parameters, and not the active parameter for MoE, didn't account for any hardware improvements etc. If they counted small based on the price or computational cost, I think they would have seen increase in small models.
[ Parent
| on HN ]
Re:State of AI: An Empirical 100T Token Study with OpenRouter
by Alex-Programs on Friday December 05, 2025 @02:57PM
(#46166464)
I think using total parameters is fair, it correlates well with the RAM prerequisites to run it. Otherwise Kimi K2 would be "small" despite being a trillion parameters!
Re:State of AI: An Empirical 100T Token Study with OpenRouter
by YetAnotherNick on Sunday December 07, 2025 @12:03PM
(#46183148)
VRAM doesn't matter if you are using API. Price and performance is what matters.