Well, you can’t. Because they are busy selling their API service and don’t have time to fork VSCode. But they will try to buy it from someone else and integrate it.
You can definitely find those here in Morocco, they would be ELATED to get an offer like that, not everyone is in the US
I posted this in another discussion thread but feel it is very relevant here.
Cursor is undoubtedly the best in the game and absolutely should be regarded as such, however I think that their business and financial side of things needs some serious help/reworking from the ground up. I’m unsure of what their day to day spending looks like, but being a paid user for almost if not a year now, there needs to be renegotiations with model providers, researching how much $$$ is spent per user per month, and tracking and understanding the patterns between the software, the financials, and the community. At this point in time it seems like they cannot figure out how to combine all 3 to make a seamless user experience that is built with a positive community. Throughout 2025 a lot of users trust has been broken due to them going back on promises and making the community and user experience worse. All we can do is hope things get better and prices get cheaper, but we’ve been hoping that for a long time now.
1 Like
Hey @jrista im trying to find the Pro+ you are talking about
the docs say
Based on our usage data, limits are roughly equivalent to the following for a median user:
-
Pro: ~225 Sonnet 4 requests, ~550 Gemini requests, or ~500 GPT 5 requests
-
Pro+: ~675 Sonnet 4 requests, ~1,650 Gemini requests, or ~1,500 GPT 5 requests
-
Ultra: ~4,500 Sonnet 4 requests, ~11,000 Gemini requests, or ~10,000 GPT 5 requests
but the website pricing page says three are only three (well teams and custom are below it but talking indiviaully)
Yeah it think this is spot on. Probably the part the pro-cursor brigade seem to just say “meh whatever” to.
you can find it in manage subscription in cursor.com your account
or in cursor app when you hit your limit and try other request it let you option to upgrade to pro+
It might be less on Cursor and more that the cost of AI is itself volatile, and having competition pop up constantly over the year is making it probably hard to navigate. I agree, Cursor appears to be the best in the game.
Merely activate the project’s automatic summarization feature, and everything will become more economical.
@ruihuyan summarization is active.
@MidnightOak I would agree.
@putt Our team is already doing negotiations, at our scale it wouldn’t make sense if they weren’t. At any rate you might find elsewhere more info about AI providers pricing changes as I won’t go into that.
Jesus Christ fam, you input about TWO MILLION tokens to claude-4-sonnet-thinking, one of the more expensive models.
That many tokens is roughly equivalent to 100,000 lines of code. IDK how you’re prompting your AI but that’s your problem.
o3 is a more expensive one too. Users like you are why we can’t have 500 request plan anymore 
Write more efficient prompts, use a cheaper model when possible, and be amazed at how cheap using Cursor actually is. From my dashboard:
This is from 40 requests, and at least half of them have been from fairly long-running chats. At this rate I’m on track to get in 500 requests and still not be billed any extra than my base $20.
Good lord, all these users on here aggressively over-using the most expensive models, and then complaining non-stop about how it’s all Cursor’s fault. I hope you do all leave to competing products and leave the rest of us in peace.
2 Likes
Hi! How is this possible?
What 100,000 lines of code are we talking about if tool usage counts toward the total? Even working on a small Python microservice can consume 100,000–2,000,000 tokens in a single request, while the agent might only produce around 150 lines of code across several files. And that’s without Max Mode.
1 Like
Hi.
On this topic, here is what I do, which it seems to me is both cost-effective and code-effective.
I pay for Cusor Pro, 20/month, fine– may switch to yearly (16/month) after a while.
I start the month using Claude 3.5 Haiku (premium)– which (AFAICT) has the slowest credit burn rate for all the premium models.
If I run out of credits, I switch the model to Deep Seek (free) for the rest of the month.
My use case is production dev work, low to medium complexity, lots of refactoring etc, full stack, .net, Sql Server and Oracle, etc.
It works (for me at least) and saves me time– these are models are quick and sufficient, for me.
This is just one way to do it.
HTH.
Thank you.
– Mark
1 Like
What do you mean you use deepseek for free? I see these in the cursor model selections, are these free?
Yes, it is something like that, so if you have cursor pro, and use up your credits for a given month, then you can switch the model to deepseek-v3 and use it for free until your credits renew when the next month starts. Something like that. I am no expert, as I am new to cursor pro, but I did some searching on it and that seems to be the case. AFAICT.
1 Like
well not anymore, claude code ever since the service outages 2 weeks ago have reduced their model capability. Opus4.1 is nearly the same as Sonnet4.
Anthropic is a scam company, they are doing this all the time going back at least 1 year now since i’ve been using them. They charge you the same but degrade the model capability without informing their customers.
1 Like
i’d recommend using gtp5-mini, much more capable than any claude model in fact (even Opus4.1)
2 Likes
Last month I used $94, this month I only used $43 to trigger the limits
Appears to be free after a few requests.
Update:
Or rather very cheap
Claude-4 is indeed expensive to use. Sure, it’s very good and helpful when starting a new project with extensive summaries, etc., but at some point it stops following instructions and generates lots of tokens in optimistic summaries even for simple edits that don’t work. I’ve started switching between GPT-5 and Claude-4 as I’ve learned how both behave. I hope more models will arrive and push Anthropic to lower prices.
Cursor has no such plan. If there were, I would buy it even for five hundred dollars.
There’s no such thing
You don’t need to prove it it really is the most expensive right now!
There’s no debate about that, but it delivers the best results.
The question is whether you’re looking for cheap or for quality. As a programmer, I’m looking for something that gets the job done quickly and saves me money.