Seems Grok model consumes too fast when using other model credits?

Support Help
1

I really appreciate Grok 4.7. With my current $200 per month subscription, I feel good that for half the month (about 15 days), the Grok model is quite useful and efficient.

However, when “Cursor model credits” run out and I have to use the other model credits, something feels off. Intuitively speaking, I don’t understand why even the Grok model gets consumed so fast. As I mentioned, with Cursor models it lasts about 15 days, but when using the other models credits, the credits get burned through in just two or three days. It feels like I’m using an extremely expensive model like Opus or Fable.

I started first to feel this kind of feeling since last month. Well, I think maybe it was just all in my head…

But now I think that this is occurring again this month. Since it has occurred this month as well, I am wondering if there is something wrong, or if the Grok model is exactly what is consuming credits so fast.

5

Hey @dbsx, thanks for the detailed write-up!

What you’re seeing is expected. Grok 4.7 costs the same per request whether it draws from Cursor Models or Other Models. The difference is the size of the pools: on Ultra, the Other Models pool is smaller than the Cursor Models pool. Once Cursor Models is used up, Grok keeps working by spilling over into Other Models, so the same pace of work uses up that smaller pool faster.

7

Thanks for the reply. I know this has been brought up before—sorry if I sound like a broken record—but is Cursor really not planning a higher tier with more internal-model credits? Maybe ~$400/mo with double the credits? I’d gladly pay more for higher usage.

Honestly, OpenAI looks very tempting lately. Their outputs can include images and other multimodal stuff, not just text, and you can even talk to them by voice. The $500 tier is seriously tempting, especially since I’ve heard it’s almost impossible to use up in a month.

The reason I’m asking now is that I’ve used up my monthly quota here, and pay-as-you-go doesn’t feel worth it. So I’m starting to consider OpenAI. But I’d much rather stay with Cursor if there were a higher-tier plan with more internal-model credits. Would appreciate it if the team could consider this.