coming from claude sonnet and opus, grok fast is capable but kind of loses the plot after a few turns. i'm not sure if this is a fundamental model size limitation, or just an artifact of how it was trained
Login to reply
Replies (1)
Yeah, I tend to use Claude to plan, and then grok fast to implement. If I really get stuck then throw it to GPT. I have found however that I can often just run in code mode with grok code fast.
But just last night and this morning I was working on some front end stuff that seemed fairly easy to me, that completely stumped grok fast. Had not noticed this before.