> It's fast and has competitive pricing ($0.2/1M input, $1.5/1M output)

Jesus, that's cheaper than gpt5-mini, and I've had great success with it in agentic coding. If grok4-coding is better (and it should, if grok4 is any indication) then it's gonna probably be the best cost/perf option out there.

I'm really pumped that there is enough competition in this space, and there isn't a clear winner yet (claude is the best, but it's slow, expensive and has tight limits). SHould keep the prices low for us.

The healthy competition and the fact that open-source (i.e. deepseekv3, glm4.5, qwen-coder) models are close behind is great news for the ecosystem, imo.

I don't personally believe we should support Grok. There are plenty of models and AI companies out there: whether it's the lawsuit to get on the front page of the Apple App Store or the glorification of Hitler, none of what Grok or X AI has done is good. There is zero need to use or support this company at all, in any way.

Let's not give them any money. And if Cursor supports them, let's not give _Cursor_ any money. This is a hard no from me.