What to do now? No more claude 4

Hey, I use a large 9k codebase. Only claude 4 is able to fix and code in it. All other models can’t fix anything due to the size.

Now 12days into my sub, I used my claude 4 limit and got same messages as everyone.

is there another product/way to not get limited for claude 4?
this is all vibe coding btw

3 Likes

Yeah, same here.

Everything was fine until the 1.2 update. Before that, I could use Claude 4 a lot—like 500 requests per day with no problem (no abuse). Then after the update, I suddenly got rate limited after maybe 40 requests in a day.

That happened earlier today (June 4), and so far nothing has reset. Even though the docs say it should clear in “a few hours,” it’s already been 10 hours and still locked. Overall, the info about rate limits and usage is just really confusing.

I also tried switching to Claude-3.7-Sonnet, and then it started telling me:

“You’ve hit your free requests limit. Upgrade to Pro…”

Like… I’m on Pro plan?? Hello??

4 Likes

I have the exact same thing. :fearful: :locked:

I had like 7 Agents prompts and reached the limit right again!


Switched to 3.7 for another 5-6 agent prompts and hgit the limit again.

This cant be right

1 Like

What does 9k codebase means? Mine is much bigger and no issues. Is it spaghetti code or modules? If you are having size issues you would hit super strange bugs and problems if you don’t break it properly with any model soon.

1 Like

one html file, in it html, css and js - approx 9k lines.
claude 4 can handle it easy…

@Mr.AB its not recommended to have that much code in a single file. Its not a codebase :slight_smile:. That Claude 4 (which model) can handle it is actually a bad sign. because humans cant handle that!

You may need to split the file into logical parts.

There are principles like SOLID, DRY,… ask AI about it to help you with this.

yup, you’re right, I always used to code that way. What happened is I was testing AI first time ever, and before I knew it I was 6k in one file, so just kept going. next project definitely will split.

2 Likes

The issue it creates:

  • Its 9000 lines of code that has to be read by AI every time so it knows what to change.
  • Repeated handling of such a file consumes more tokens and wastes your Pro plan.
1 Like

Thank you @T1000. I’m actually splitting the file rn

2 Likes

Anyway problem is still exists. It takes a lot of token even on small files.

I know what you are talking about, tested it myself, can confirm it can handle it, but can also confirm that it is bad in a context of results that you get. Since it is done by Cursor, I am sure you have comments for different modals in there. Ask it to extract those modals into separate components in separate files, and your life will get much easier, also when it comes to expenses, especially now. More lines more context more money and bigger chance for something to go south (hint: ask it to localise your file in 2-3 languages and see what happens).

1 Like

@FlamesONE I am not aware of which issue you are talking about. this thread is about OP having a specific issue due to large file. I recommend filing a detailed bug report as a separate thread with the issue you are having so it can be addressed.

Missing the 500 package a lot

3 Likes

Almost 19 hours later and model is still rate limited… there is definitely a problem or cursor are rate limiting harder.

3 Likes

@erophames @FlamesONE I wont start here discussions that are out of topic here or have their own threads. Thank you for your understanding. Please respect forum guidelines.

4 Likes

I understand, just give a answer in new thread about this problem

4 Likes

you need pro plus now apparently lol.

yes, I upgraded to Pro+ and gained immediate access to all models. Not sure when I will hit “rate limit”…