MiMo 2.6 Pro is at the Pareto frontier (the one where you only need 20% of the smarts for 80% of the tasks) according to Artificial Analysis, nicely filling in as a substitute for a hypothetical 'GPT-6 Terra' (which doesn't exist as far as we know). That's pretty darn impressive from an open model.
That's not what the Pareto frontier is; you're mixing up Pareto frontier with Pareto principle.
https://en.wikipedia.org/wiki/Pareto_front
https://en.wikipedia.org/wiki/Pareto_principle
Despite the error in the parenthesis, it's exactly what he says: https://artificialanalysis.ai/?intelligence-category=text-on...
I was just responding to the error in the parenthesis.