Comment by mehdibl

What matter is not context or the recod token/s you get.

But the quality for the model. And it seem Grok pushing the wrong metrics again, after launching fast.