Please quit
Sorry, as you say in the banner, limited to pay users.
However, is is impossible to use claude 3.7. I would suggest not announcing that you are offering it if paid user are just able to access it once every 20 tries. It is just a fraud.
1 Like
Exactly Cannot use that model because if high demand
It’s called bugs, I was having issues until the last patch but that has mostly resolved. I have found that switching to open ai model for a couple messages then switch to 3.7 seems to knock it loose. I think it would be great if they offered like a canary program so people that signed up for it would get the update At the same rate, we’re getting them now and others could choose to wait until cursor deems the release stable.
Also, you can’t ever know how a large volume of users are going to impact Performance I think that in many cases anthropic’s API is rate limiting because of the high demand not just from cursor, but across the board with many AI application applications.
Exactly that. Completely agree.
Main userbase should not be affected either by Cursors IDE updates AND by Cursors supplier Anthropic doing their updates. Thats why there are versions of models, SLAs etc in B2B connections/dependencies like Cursor and Antropic.
Bugs are understandable, making a working system so hard to use that users have no other choice but to STOP is not just a another ‘bug’ situation that.
1 Like
For Claude, the token count of 20k per call is what makes it not work when using your own API key. I couldn’t get the thinking model or the normal model of 3.7 to work as it just hits the token limit and says I’ve reach my limit. In fact, I’ve only used $0.15 per call, but the tool call via my own API has a threshold that Cursor breaks every time.
I don’t have o1-pro so cant comment, however it is well behind Claude for coding capacity according to https://web.lmarena.ai/leaderboard
Seems to work great in first test… Perfect!
I think it would be great if cursor had something like a canary release so people could choose to be on a separate release schedule.
I imagine you probably already have some testers that get early access but that may not help test super high volumes.
I expect that many people would choose to be on the Canary release that gets announced but having the ability to easily switch off of that release schedule to the last stable implementation would be super helpful for lowering the volume of duplicate issues that the cursor team receives. Which I can only imagine, adds significant workload and slows down there ability to actually fix the issues that people are reporting.
I understand your frustration with Claude 3.5 struggling with large codebases. Based on the transcript you shared, I can provide some helpful advice on how to use PRD (Product Requirement Document) and RFC (Request for Comments) approaches with Cursor to manage large projects more effectively.
Using PRD and RFC Approach with Cursor
The approach demonstrated in the Turkish YouTube transcript is quite powerful for working with large codebases. Here’s how you can implement it:
1. Create a PRD Document
Start by creating a clear product requirements document that outlines:
- What your application does
- Required technologies
- Design patterns
- Technical specifications
2. Establish Cursor Rules
Create a set of rules for Cursor to follow when working with your code:
- Coding standards
- File/folder structure conventions
- Naming conventions
- Architecture principles
3. Create a Features Document
Break down your application into discrete features with clear specifications for each.
4. Generate RFCs for Individual Features
For each feature or change you want to implement, create an RFC document that includes:
- Technical design
- Required endpoints/functions
- Data models
- Database schema changes
- Security considerations
- Dependencies
5. Implementation One at a Time
Instead of asking Cursor to understand and modify your entire codebase at once, ask it to implement or modify just one specific RFC at a time.
Why This Works Better
The key insight from the transcript is that by breaking down your project into smaller, well-documented chunks, you give Cursor a clearer context that prevents it from getting lost in the complexity of your entire codebase. Each RFC provides focused guidance for a specific task without requiring Cursor to understand the entire system at once.
Example Prompt Pattern
Based on this RFC document:
[Insert your RFC details here]
Please implement/modify the following specific feature in my codebase:
[Describe exactly what needs to be changed]
Here are the relevant files:
[Include only the necessary code files for this specific change]
This approach guides Cursor’s work within clear boundaries, significantly reducing the chance of it hallucinating or making unwanted changes to your code.
Regarding your question about Claude 3.7 returning online, I don’t have specific information about when it will be back and working properly. I would recommend checking Anthropic’s official channels for the most up-to-date information on Claude 3.7’s availability.
##Resource
1 Like
Same horrible experience. I am a paying customer. First got slowed down and now completely blocked. As a developer, I look for a stable and reliable environment to conduct my work. This is completely unacceptable. I hope they get to the bottom of this issue promptly.
1 Like
Hello, I have a question: at the moment access to Claude 3.7 for free users is limited. As far as I understand, users with PRO subscription can use Claude 3.7 with fast 500 pools. Is Claude 3.7 now available for slow pools with PRO subscription?
1 Like
Some thoughts from the weekend, Dan .
The 4.5 preview is incredibly powerful, however, at $2/request it would be nice if Cursor could help warm up the prompt a bit. There have been multiple times when I’ve had to tell GPT 4.5 to proceed with an update for another $2, which can take me to $6 or even $8 a request for what should have cost $2. It’s an extremely clean coder and doesn’t try to impress me with pointless architecture the way Claude 3.7 does but Cursor could do a LOT more to ensure I’m not being double, triple, or quad charged for the pleasure.
Also, the latest Cursor on Sequoia still experiences a slow entropy towards VSCode error 5, to the point where I’ve considered putting a pnpm clean:vscode shortcut to “m -rf ~/Library/Application\ Support/Cursor/User/workspaceStorage” in my application. It’s really bad. Manageable now that I know what is happening but still pretty shocking given what I pay Cursor per month.
I was hit with a surprising $34 bill today. I simply booted up my computer and started coding like I do everyday. Looks like there’s a new model that charges $2/request which I didn’t select but somehow the IDE defaulted to it. I have emailed [email protected] about this issue. Please reply at your earliest convenience. Thank you.
For others that are reading this message, double check which model you have selected or even remove it as an option from your account to prevent this from happening to you too.
What is the rate limit of using Claude 3.7 thinking? Meaning, how many prompts are allowed per day, for paying users.
1 Like
The Cursor agent/editor has been essentially useless in our workflow. It consistently fails to solve even straightforward development tasks, while a simple direct call to Claude 3.7 resolves the same problems immediately in a single turn. This poor performance significantly undermines any incentive to continue using Cursor. I suggest prioritizing improvements in reliability and effectiveness to make it genuinely beneficial for developers.
1 Like
Why is 3.7 still inaccessible for trial and free users?
There’s no reason for it. status.anthropic.com shows everything is fine and status.cursor.com as well.
If you want to hide sonnet 3.7 behind the paid plan only, stop being disingenuous and tell us rather than using a fake excuse of “try again in a few minutes”
actually you should be happy since 3.7 is not available, as a paid user, i have disabled 3.7. the current version is useless, even so, cursor’s pride tab feature still eating texts and insert tab instead of bring of prompted text.
I’ve tried sonnet 3.7 when it was available (just after release), you need to have it under a VERY VERY tight leash.
I want 3.7 reasoning for the thinking part. I wouldn’t let it code 
Still mesmerizing that the cursor team is showcasing (useless) features such as using MCP to order Uber Eats and working on future releases such as custom agents when their integrations of current models is less than sufficient.
It might be available but it doesnt work!
If you ask Claude 3.7 to just fix a bug, he’ll fix you a bug, add features you don’t need, order a pizza, walk the dog, etc.
3 Likes