Are the model performance metrics introduced in the blog for the Base model or the Chat model?

[underspirit]

This model card mentions that grok-1 has undergone fine-tuning: "The model was then fine-tuned using extensive feedback from both humans and the early Grok-0 models. "

But, this blog mentions that grok-1 is not fine-tuned: "Base model trained on a large amount of text data, not fine-tuned for any particular task."

That is confused, Are the performance metrics displayed here for the Base model or the Chat model?

https://x.ai/blog/grok