Where has it been moved to?

Previously, the remaining token value used to be displayed in the image position, but now I can’t see it. Where has it been moved to?

Congratulations, you’ve been moved over to the new pricing plan, which has no concept of fast requests - they’re all slow now, at least Sonnet requests are.

If you want those tasty fast requests back, open settings → advanced settings and opt-out from the new pricing model. Just make sure not to hit “delete account” right below.

1 Like

What has improved? Does this mean Sonnet 4.0 can be used without token limits? I don’t understand the pros and cons of the old pricing model versus the new one.

taking forever to load sonnet request now. nice upgrade team :smile:

1 Like

Kind of it can.

In old pricing model:

  • You had 500 fast requests (and most of the time they indeed were fast)
  • You had unlimited slow requests with no internal rate limits (but not for Claude 4)

In new pricing model:

  • Yes, we can use more than 500 Sonnet 4 requests, yay!
  • You no longer have guaranteed fast requests that you can rush through when you need them
  • You still have so called unlimited requests that in theory shouldn’t be slow (meaning artificially slowed down like in old pricing model), but people complain about Sonnet being super slow now
  • Those unlimited requests are still rate limited, so you can only do as many of them over a few hours window
  • We don’t know the details of how many requests and what requests lead to rate limiting, but we know you can switch to some other (weaker) model and hopefully not be rate limited anymore
  • Using longer prompts or conversations leads to getting rate limited faster
  • In theory it’s designed to be better for most users, in practice - we’ll see

From my POV, as I have a preference for longer conversations, Gemini on old Pro was a good choice. I’ll stick to old pricing model until I use up my fast requests and then try the new pricing model to see how rate limits behave.

Check the new chapter in the docs for some (vague but still) details: Cursor – Rate Limits

1 Like

thx for reply

Yeah gustojs described it well.

However note that Anthropic has/had today outages with Sonnet so this is why Sonnet is slow or not working sometimes.

1 Like