I need a monthly subscription to use UNLIMITED fast requests. I am willing to pay a higher price than the Pro plan. I am a user who can take advantage of the AI power very well in Cursor so I use AI often, so I do not want to use the “usage-based pricing” function, I will definitely go bankrupt because the number of requests per day is not small.
The solution for this is to just diable usage-based pricing, isnt it?
I dont use it but i remember it was a feature that could be enabled/disabled.
Just enable the usage-based pricing and usage-based pricing for premium models … Taaaddaa ! lol You are charged 20$ per 500x Fast Requests so … you wont go bankrupt … maybe if you use MAX models but thats your choice …
I completely agree with this sentiment. It would be fantastic if a monthly subscription plan offering unlimited fast requests at a reasonable price were introduced. Such a plan would allow many users to enjoy the service without worrying about usage limits, and it could greatly enhance the overall user experience. I sincerely hope that the developers consider launching an affordable, unlimited plan soon.
They cant … models aren’t unlimited … its not cursor that make the limits its the AI models/company … Just get usage-based pricing and youll be charged 20$ per 500 fast requests … Unlimited if you want but 20$ per 500x Requests … lol its not hard to understand ^^
With usage based pricing after the included 500 fast requessts you pay per request at same rate like the subscription charge but not in 500 steps, just as much as you use.
Once AI becomes a commodity then Unlimited will be reasonable as for now each request is handled by API and charged separately.
I experience this with internet acces fees, first per hour, then per MB, then per Mbps, then per GB/month at top speed. Some providers offered unlimited but after x GB capped speed or kicked users out when they consumed so much traffic e.g. downloading half of internet lol. Now its less of an issue.
I agree that in whole AI field the per token and per request pricing is not practical but due to still high cost of such servers, Google/Anthropic/OpenAI - who run the models which we users select in Cursor - charge per token. For one im glad not to pay Cursor per token as often my detailed Agent usage goes through decent token amount and still allows me 25 tool usages (which does trigger internally multiple provider API calls for sure and not just API call per 1 request in my use case)
How much are you willing to pay? Your “need” isn’t helpful without a qualifier. Tell them here what you are willing to pay. $500? $1200 $2300?
Anyone can set their limit in Usage Based pricing . Start small and increase.
While some have heavy usage, for many its < 20 or 50 usd.
We have thought about this before, but as we have a direct cost per each LLM request, an “unlimited” plan could only be truly unlimited if it was crazy expensive outright.
With usage-based pricing, you’ll only ever pay for what you use, and as @T1000 has mentioned, you can set an upper limit to ensure you don’t use more than intended.
Sorry if I made you angry. I just thought Cursor has some special collaboration with AI companies so they can give a great price for using “somewhat fast” requests. Actually it doesn’t need to be “super fast” I just need a request of about “20 seconds for 1 response”. If there was a service package like that. For example 3000 “somewhat fast” requests for 1 month, that would be great, right everyone? @T1000 @danperks
The reason I want this is because lately claude 3.7 sonnet is really slow for 1 response. Probably about 2-3 minutes or more. Really annoying :((
No worries, didnt get angry at all but meant that with time AI will become a commodity where its so common that it will become basically a requirement and cheaper, like internet nowadays.
We ALL would like super fast AI APIs and that 3rd party providers like Anthropic dont make software using their API broken as we more and more rely on them and have prompts tailored.
(im also just a user)
Cheers
@DemonVN While we dont have a $50/m plan available, you’d be able to get 1250 premium requests for that, using the 500 included in the $20/m plan, plus $30 of usage based pricing charges!