Are you using the free version?
No! The top tier paid version!!
Today barely anything is working!!
You mean access to Claude 3.7?
No. Claude3.7 doesnt even work! It is just permanently Overloaded!
Claude 3.5, today is now completely and utterly USELESS at everything! Yesterday it was working really well. The best i have experienced it actually! (3.5, not 3.7). But today, it can’t do anything right!!
I have been trying to make the most simple of code changes and I have had to reject its response 20+ times because it just does whatever it wants!!
I also tried reinstalling the software, but now that has new problems. Search bar has disapeared! Composer has gone and been replaced by this ornamental pile of garbage!
i dont know how you guys feel.
but i kept stay on 0.45.8, i can still use agent mode and C.37t.
everything is fine, cursor IDE is working for me.
yeah sometimes C3.7t not available, no matter pay or not,
yet, this setup works, for me.
1 Like
I switched to sonnet-3-7-thinking for everything several days ago, and it is much better that 3.5.
It even created me a Chrome Extension to convert currently opened Confluence page to Markdown (and copy to clipboard) with 1 shot!
I spent more time generating an icon for that extension longer than everything else.
It just worked. From the first time.
In last 2 hours I’ve created Product Requirement Doc, Architecture, Story planning and actual implementation for 2 stories in 7-story epic. With unit and E2E tests.
Yeah, there were 4 or 5 times errors like ‘Anthropic is busy’ but retrying helps.
Idk what else to say here.
Try restarting Cursor when you see some weird behavior. Helped me several times already.
Prey that the release stabilizes over time.
Available? Not really, rate limited even with a fresh amount of fast requests
I just installed and made pro account. I don’t have 3.7 as an option for models!? Any ideas?
1 Like
It DOESN’T work though!!
Permanently in High Load default mode all day today, and other users reporting this for the last few days also!!
Pinning posts like this to the top of the forum is like rubbing salt in the wounds while kicking you in the head!
2 Likes
you do everything except fix Azure’s o1 and o3 models. Do it finally. It’s a simple task
I don’t understand why, but almost 80% of requests to 3.7 go to the trash, it searches something on the Internet, something reads in files, and then everything disappears as if it did nothing. Has anyone encountered this problem?
Can we get any sort of credit or refund for GPT 4.5?
While I see the pricing in the latest update, when I ran the requests there was nothing in app to indicate it would be 5x the cost of the next closest model… absolutely was NOT worth it.
PUMPED!!! Thank you! Appreciate your incredibly fast addition.
Yes its possible you can use it with slow request but this exceptions:
*If 3.7 sonnet using high volume, only pro users use this model with prority.
Thank you very much, if you and others interested how to do it PRD and RFC documents for Cursor ai, i can recommend english youtube video for community.
i think its not possible right now, many users have complaits about that issue…
Hey @khallmark @mk148a,
If you were unaware of the increased costs, please drop us an email at [email protected] and we’ll be happy to help you out with this!
We hope to improve on the visibility of high-cost models soon.
I totally get your frustration — it’s disappointing to hit usage limits when you’re a paying subscriber, especially if you’ve been consistently active and supportive over time. The idea of “unlimited” should mean just that, without hidden throttling or priority restrictions based on usage history.
High server load is understandable at times, but it would be more fair if they were transparent about how limits are applied or offered a more balanced approach (like usage tiers or prioritized queues). Hopefully, they listen to this kind of feedback and adjust their policies accordingly.
You’re not alone — a lot of us are keeping an eye out for a platform that gets it right.