If you were in charge of pricing at Cursor, what would you do?

Discussions
1

Discussion on Cursor’s New Pricing

Many people are finding it difficult to adjust to Cursor’s new pricing plans, especially now that the AUTO mode is also becoming a paid feature.
Right now, in the near future, users are going to find themselves just a few days after the start of the month unable to use any model!
That means that even for a paying subscriber, Cursor will become unusable very quickly.

I’m curious if you were in charge and wanted to keep the company profitable, how would you approach this?
How would you price the usage?

On one hand, you have to deal with API costs.
On the other hand, you need to keep customers happy.

Possible Ideas

Personally, I think I might:

  • Negotiate tailored deals with certain companies.
  • Offer open models for a fixed price.

But I’m not sure how practical that would be.

1 Like
2

An open model fixed price tier is a very good idea, imho. That can be set up (custom models, or third party extensions), but it would be nice if it was a built-in option

The difficulty with paid models is this is a unique product in which you don’t know how much it will cost (what the model does) until after you buy it, e.g. you do not know that what you are about to ask will cost $1.75. One of the most important features of a healthy free market is price transparency, so consumers can make informed decisions and exert healthy market pressure. This is not Cursor’s fault (at least not the root of it), the root is that the models can not provide predictive pricing. I am sure Cursor would love to offer the best pricing available because the competition is so fierce in this space right now….so that is their challenge, big challenge

3 Likes
4

I would probably adjust the api markup from 20% markup to maybe a 10% markup?

People are already paying a monthly charge in most cases, so hopefully they can be profitable on the monthly included credits and then any additional usage based pricing could be less of a markup OR a reduced markup solely for people who have paid monthly subscriptions could be a much easier pill to swallow for most.

Just a thought…

20% markup for solely usage based users.

10% markup for users with paid monthly subscriptions that have the usage based billing enabled.

Maybe an even lower % markup for ULTRA users? If they hit their limits, they are clearly using a lot and a 5-7% markup would still likely yield a good amount of profit due to their volume.

Just my thought :person_shrugging:

5

They just demonstrated that you can’t build a business selling something $0.25 that you pay $1.

4 Likes
6

An open model fixed price tier is a very good idea, imho. That can be set up (custom models, or third party extensions), but it would be nice if it was a built-in option

The difficulty with paid models is this is a unique product in which you don’t know how much it will cost (what the model does) until after you buy it, e.g. you do not know that what you are about to ask will cost $1.75. One of the most important features of a healthy free market is price transparency, so consumers can make informed decisions and exert healthy market pressure. This is not Cursor’s fault (at least not the root of it), the root is that the models can not provide predictive pricing. I am sure Cursor would love to offer the best pricing available because the competition is so fierce in this space right now….so that is their challenge, big challenge
[/quote]
You’re right, but it can’t be denied that they’re currently the only ones working based on tokens and not requests, right?

7

That’s a really great idea, and it looks like they’re already implementing it. I want an idea they haven’t implemented yet. :thinking::face_with_raised_eyebrow:

8

I’m not sure about what you’re saying — are you certain they’re taking the twenty percent from you?
As far as I know, Cursor actually provides the API at a cheaper rate than the companies themselves, and they make their profit through internal agreements with the providers. I recommend you double-check this point.
Also, your suggestions still don’t allow for pricing that would enable continuous use of the software.
Right now, with Cursor removing AUTO, there will be users who suddenly won’t be able to use the software at all with any model!

1 Like
9

yeah, good point and idk how many of the other companies are profitable, or are they just burning through their funds.

10

I don’t really understand so what are you suggesting?
I suggested a solution: making deals. There are many providers willing to offer a monthly API plan to such companies at a fixed price, so it could at least be profitable for Ultra subscribers.

11

You know, or this is what they are repeating on a forum? Seriously asking.

12

I would first figure out what is my product and who is my competition. Currently I am not sure if they are competing with OpenRouter or Windsurf or whatever.

In a world of software development where developers mostly believe in Claude, they need to offer either same price of what Anthropic is offering with Claude Code (which is currently even not comparable) or better quality. Unfortunately the quality is worse, and price is higher (I am really not buying kids stories about giving 50$ for 20$). But this all goes back to - what is the product that competes? If this is Cursor with all its bugs, chaotic release strategy with mass QA by people who actually pay for it and get charged for every request not knowing is it normal charge or again some bug somewhere - there is no pricing strategy for that.

And let’s not forget OSS LLMs which are coming and which are better and better every couple of months. And as free as your local machine can take it.

13

Which detail are you asking this about?

That’s a very true and very sad point.

I don’t see how LLMs running on the computer would be able to help you.
The main problem is never writing new code — it’s fixing bugs and locating the right place in the project to add code. That’s always the tricky part, and what you wrote would help less with that.

14

I’d go find another job. Cursor is washed.

15

My understanding is that Cursor used to pad the LLM providers costs by 20% when using MAX mode but under the new model they’re going to be basically providing the LLMs at cost (they’re adding $0.25/million tokens but that’s to cover other operating costs).

I guess this means that going forwards they’ll roughly break even for heavy Pro users, make a bit of money from those who don’t use their full monthly allowance and print money from Teams (charging $40 for $20 of LLM credits).

The biggest issue with the pricing model isn’t the cost, I don’t expect other people to subsidise my usage, it’s the uncertainty, moving from 500 requests per month on my current Teams plan I have no idea what $20 will look like, especially since you don’t show me my current token usage.

I appreciate that Cursor tried to provide consistency with the x requests per month model but I guess it could never really work as I’ve seen Background Agent tasks use 50k tokens on a backend microservice repo and > 1m on a frontend one.

3 Likes
16

:laughing:But you’d lose your salary and in growing companies like these, it could be a high salary. Are you sure?

17

What do you mean you don’t know?
Because they show you a detailed breakdown on the website.
Do you mean what I’m complaining about that you can’t see during the conversation what’s attached, etc.? Or something else?

1 Like
18

They do this for Pro but not for Teams, Teams currently still uses the 500 requests per month model so all the token numbers just show 0. They only show the token numbers for Teams when you use MAX or Background Agents.

19

I’m just an Ultra subscriber, so I have nothing to say I’m not familiar with that.

1 Like
20

I’d start with a live Q&A and lead with transparency about the challenges we’re facing as a company. I’m open with my entire team about our costs, what we’re dealing with, and why commission splits are structured the way they are. I’ve had top producers ask for significantly higher splits, but once I walk them through the math and the net tangible value, there’s no pushback because facts are facts. People respect transparency when it’s real.

That’s exactly what’s missing from Cursor. From my perspective, the lack of honesty feels systemic. There’s no transparency, none. It feels like we’re being gaslit. They sell us on promises, package it in polished word salad, but it’s all smoke and mirrors. We’re treated like we’re naive retail consumers when, in reality, many of us are highly technical, business-savvy users. The disconnect is obvious.

Cursor has publicly stated they’ve negotiated extremely competitive API pricing due to their large user base. Fine. Maybe NDAs prevent sharing exact numbers, but there has to be some way to communicate cost structure or margin transparently. As it stands, everything from them sounds like a sales pitch. I read their updates just waiting for the “gotcha” moment.

How do you rebuild trust from that? You do it with transparency. Real, measurable transparency. If these are truly your costs, and there’s a small, reasonable markup (10% maybe, 20% is too much at these rates), fine. At least then we can make an informed decision. And if Cursor is really trying to create better deals on the backend to lower pricing over time, that’s great but prove it.

Right now, it doesn’t feel like that’s happening. The new token caching system, in particular, opens the door to abuse by Cursor and users have already demonstrated how that has been exploited and exponentially increased usage costs. Add to that the frequent errors, disconnections mid-session, and the fact that we’re being billed again just to hit “continue”… I’d estimate 50% of my billing is from those kinds of disruptions alone.

Bottom line: people feel overcharged, misled, and increasingly alienated. Month by month, the reasons to stay are being eroded. Cursor needs to stop pretending this isn’t happening and start engaging with the community like we’re partners, not just revenue streams.

6 Likes
21

I must point out that this is something I see repeating itself there’s a certain lack of cooperation with customers that I don’t quite understand the cause of. It’s not just about money; it’s in many details in the way customer service works, the responses, the overall management not being organized enough. I find it hard to put it into words, but you described it well.
However, you didn’t give a solution you only pointed out that you would manage the current situation in a more transparent way, which is an interesting direction, but it doesn’t help customers who are also looking for a cheaper plan.