Cursor Paid Tiers and Upsells - Discussion

Discussions
1

I love what Cursor is trying to do and appreciate how hard it must be to try to rapidly iterate and innovate given all the competition. I’d like a thread where the community identifies the value in the product so that hopefully we can influence how Cursor team positions and builds revenue.

I had no issue going straight to $20/month.

For the last 8 weeks, I’ve been using cursor a couple hours a day and treating any limits as a good time to step back and plan or work with another agent (Claude/Gemini).

I feel like I was gifted(?) a ton of usage for Composer when it came out and it really felt like a big step (faster, easier). My usage meter basically didn’t move much at all for weeks.

Recently, usage meter took off, sometimes blowing up 10% on a single query. I wasn’t changing anything that I had done in the past. When I got past 80% I stopped using Cursor because I didn’t want to get caught in the middle of some work.

I have a long history with the Web and apps (back to 1994) and pioneered some of the very first freemium products so I do feel for the painstaking process of trying to get people to pay and still make it a product that people love.

This was the first time, I had this sentiment that I could just stop and not use the product (or come back to it at the month reset) which to me is not a great sign. So I want to get other’s feedback on what they are feeling. One of the issues of having many companies releasing at rapid pace is that it creates strong deflationary pressure (from a psychological perspective).

Thank you for all the hard work. Amazing how far we have come. I built my first websites on Win95 using notepad.exe and FTP client.

1 Like
2

I feel you, a lot of people are feeling the pain as the new usage model comes into play.

Getting introduced when they were giving away usage versus now, where most is charged… I know I am needing to be more mindful of which LLMs I am using, and what workflow steps I use to accomplish a goal in order to avoid blowing out a usage budget… Also, as a project becomes more complex, more context is needed in order to properly account for dependencies, etc. That also adds to overall context use and impacts usage burn rates. I am ok with all that, understanding the differences between using a $2 per Million token or $40 per million model is part of working with ai…

But when those issues cant be managed adequately because of “new features” that inadvertently burn through all your tokens without out your awareness.. that is where I m afraid we have to draw the line.. Otherwise the overall value proposition of Cursor begins to evaporate. Particularly when working in teams or enterprises… managers rarely have sympathy when their budgets explode and your not sure why or how to anticipate it in the future short of not using the tool to begin with ….

3

Applied rules and available tools consume tokens. Each tool call, including mcp calls paste all previous context and start over. It used to be like this, I think it still is. Most of these are recognized as cached tokens so they are cheaper. I suggest trying a blank new project, no mcps, no rules and check how much tokens it uses and in dashboard check how many of these tokens are cached.