Daily reset - when?

General
1

For some models, there are restrictions like “10 fast requests per day,” but which time zone is this based on? Does the time of the first request matter, or is the day always determined by some preset time?

It might sound nitpicky, but I’m trying to plan my work time.

Thank you.

1 Like
2

I am assuming you are referring to these models at https://www.cursor.com/settings?

  • Chat, Cmd-K, Terminal Cmd-K, and Context Chat with claude-3-opus
  • Long context chat with claude-3-haiku-200k
  • Long context chat with claude-3-sonnet-200k
  • Long context chat with claude-3-5-sonnet-200k
  • Long context chat with gemini-1.5-flash-500k
  • Long context chat with gpt-4o-128k

Just adding this info and screenshot here in case other users weren’t aware of these allowances.

settings_pricing_details
settings_pricing_details1584×842 106 KB

I suppose the daily refresh time could be added to the UI, eg:

Daily allowances refreshed at 00:00 UTC.

or:

Daily allowances refreshed at 00:00 PDT (UTC-7)

The ‘usage events’ on that page seem to display in local time, but I’m not sure if the daily usage window would be in local time or another time zone.

3

Hmm, I don’t understand the usage based pricing part. I thought that cursor pro subscription offered unlimited access to premium models and that the only difference was in how fast the responses are.

Am I understanding correctly that the usage based pricing is to get fast premium access?

4

Or in other words, I’m mostly curious in understanding whether pro subscription alone would be sufficient or whether I should opt in for usage based pricing as well. :thinking:

5

Hi @arun,

Just to re-emphasize the text that is displayed when logged in at https://www.cursor.com/settings:

You may opt in to usage-based pricing for requests that go beyond what is included in your plan. Currently only applies to Claude 3 Opus and a few models in long context chat.

So this is an ‘opt-in’ feature that Pro user’s can toggle ON or OFF.

It is OFF by default.

When toggled OFF (the default setting):

  • As a Pro user, you get 10 requests per day for these, mostly long-chat, models:

    • claude-3-opus
    • claude-3-haiku-200k
    • claude-3-sonnet-200k
    • claude-3-5-sonnet-200k
    • gemini-1.5-flash-500k
    • gpt-4o-128k

When toggled ON:

  • You can make more requests to these models, but it costs 10 - 20 cents per request.

Unless there is a specific need to be able to make additional requests to these specific models, and pay for them, I would recommend leaving it set to OFF, which is the default setting.

Also note that all of the above information is in addition to what Pro users receive, ie:

  • Unlimited completions

  • 500 fast premium requests per month

  • Unlimited slow premium requests

  • Unlimited cursor-small uses

To clarify your question about premium and non-premium models…

These are the premium models:

  • gpt-4
  • gpt-4o
  • claude-3.5-sonnet

These are the non-premium models:

  • cursor-small
  • gpt-3.5-turbo
  • gpt-4o-mini

In regards to what fast and slow uses means…

Pro users get 500 ‘fast-premium’ uses per month.

After that, their requests to ‘premium’ models are put in a queue, and therefore returned ‘slower’.

For your reference, here’s another topic that discusses what ‘fast’ and ‘slow’ means.

1 Like
6

Thank you, this is super helpful.

My confusion stems from the fact that I thought, earlier, that claude-3.5-sonnet and claude-3-5-sonnet-200k meant the same thing. It looks like they are not.

1 Like