didn't read the article, but just the title felt fishy because of course an inference provider wouldn't mind claiming to provide access to a mythos-class model.

didn't read the article, but just the title felt fishy because of course an inference provider wouldn't mind claiming to provide access to a mythos-class model. So far, GLM 5.2 is still my go-to, even at less than a third the parameter count compared to Kimi K3