Issues with performance? Unusable? Slowness? Check here and vote on fixes!

[jdubb75]

Hi Cursor AI team and community,

Summary

This post qualifies as a consumer review under FTC rules (16 CFR §465.1).

Feel free to throw a like on this post or to comment to help boost and maintain visibility, without replies it will fall to the depths of the forum and help nobody.

I’ve been a user of Cursor AI for a while and appreciate its potential, but I’ve noticed significant performance issues, such as slow API requests and general platform challenges, especially in recent weeks. These issues have impacted my workflow, and I’ve seen similar feedback from others in the community.

I’ve also observed that some posts discussing these issues are hard to find in the forum, which raises concerns about feedback visibility. Open feedback is crucial for improving the product and helping users make informed decisions. If posts are not visible, it could unintentionally mislead potential users about Cursor’s performance, which might affect consumer trust per feedback guidelines.

This is a very much more recent issue, and searching through the forum shows that.

To address this constructively, I suggest:

Creating a pinned post in the feedback and bug report section to consolidate reports of slowness or performance challenges, with clear debugging guidance (e.g., how to submit logs or troubleshoot).

Ensuring negative feedback is visible alongside positive comments to maintain a balanced view of user experiences.

Clarifying moderation policies to confirm that legitimate feedback is fully visible.

If this post is filtered out of visibility or removed, it would unfortunately strengthen concerns about feedback visibility. I hope the team can address these performance issues and provide clearer channels for feedback to benefit the entire community.

Thanks for considering this, and I look forward to any updates or guidance from the team!

• Jdubb75

Experiencing these issues and believe the forums need to be updated to address this better?

• Yes
• No

Have you found it hard to see your feedback posts in the forum?

• Yes
• No

Was this post easy to find and still visible in the forum under its category?

• Yes
• No

The below posts cannot even be searched for and have all been removed from visibility or suppressed, yet they would all be considered user reviews and feedback about the platform (potentially causing FTC issues). This shows clear, consistent suppression and a lack of guidance when it comes to these cases that users are looking for help with - whether or not it is viewed as complaining or conspiracies by moderation.

My Recent Post about slow requests disappeared — Bug or Moderation Issue?

Hi all, Yesterday, I published a post about slow requests (link: Slow requests now stuck), but now I’ve noticed it has disappeared from both the search and the list of posts. Is this a bug with the forum, or has my post been removed by moderators? If it’s the latter, is there a reason for hiding discussions about new API restrictions? Would appreciate clarification. Thanks!

Slow requests now stuck

Hey everyone, I wanted to share my recent experience and see if anyone else is running into the same issue or has a fix. Previously, I was making “slow” (i.e., free/low-priority) requests to some of the top AI models (Claude-3.7, 3.5, Gemini), and honestly, they worked almost as well as the paid options. Sure, you’d have to wait a bit longer, but you’d eventually get your answer. Suddenly, that’s changed for me. No matter which of these top models I’m trying (3.7, 3.5, Gemini), I’m now gettin…

Why is everything so slow

So all of a sudden cursor just became unusable for me a few hours ago, Everything was fine now all of a sudden not a single prompt is working, either that or the request is taking so long that I’m cancelling it early. Either way what am I paying $20 a month for, feels like every week they’re squeezing more money out of me and the products getting worse. Also i blew through $50 worth of max requests in like 2 days.I can see myself considering an alternative to cursor at this rate.

Agent doesnt seem to be agenting anymore

Very often recently I’ll tell Agent to do something and it merely acts like ASK mode. If I explicitly choose a model such as Gemini then its more likely that Agent will actually do the work, but if I leave Agent set as “auto” I get a response like this. Just a suggestion on the type of thing I might do. It’s not really worth the money I am paying ! [image]

O3 is utterly useless

tried one prompt with o3 as i had multiple tasks for the llm, o3 completed maybe 5% of what i told it and im being generous as it was a wrong solution for a rather simple php problem, ignored other instructions and tasks, consumed around 90K tokens, cost me 50 requests, tried to use python command that does not exist, let alone in a php repository, failed to use mcp server none of these problems happen with other models and before 0.50 o3 was rather superior, i used it frequently for the exact …

If you have posts that you have found to have been suppressed and are no longer visible in the forum that you feel were wrongly suppressed similar to these situations; please feel free to submit them to be added to the list.

This is not a post looking for legal action or anything in that regard; it is to shine light upon potential issues and look for building a stronger community. Clarifying due to the FTC references*

Common platform complaints that go unaddressed (feel free to submit more and I will consolidate here):

1. Chat History Continuity and Resumptions (Relying on SpecStory as an extension and having to manually back up chats to prevent catastrophic loss for users)
   Seen here, along with numerous posts - this is mainly to assist in getting attention on the topic:
   Historic chats no longer accessible after update - #3 by jdubb75
2. Auto-Run/Yolo Mode not working (added in case the initial users concern is swept under the rug, only solution for some users has been clearing the allowList but this is not a fix for all users):
   Auto-run doesn’t work - #10 by yaireo
3. Ongoing complaints about version 0.50+ that are all over the place. A lot of users are experiencing severe slowness, model performance deterioration, oddities within reasoning chains, and rules being fully ignored by models (I am working on a temporary RuleMCP to see if this can improve the rules ignoring issue for some users and will provide it if it seems beneficial):
   Cursor 0.50.5 is a mess for me - #4 by deanrie
   My Experience with CURSOR – Not Just Slow, Practically Unusable Lately
4. Many users complaining about Gemini 2.5 Pro becoming completely unusable:
   Gemini 2.5 pro 05-06: strange hallucination issue
5. Reply to contribute more to the list.

[Legit_Welsh]

I’m replying to help boost visibility.

[jdubb75]

thanks for the contribution!

[jdubb75]

Updated and added more detailed sections, for example.
Common platform complaints that have been going unaddressed (feel free to submit more and I will consolidate here):

1. Chat History Continuity and Resumptions (Relying on SpecStory as an extension and having to manually back up chats to prevent catastrophic loss for users)
   Seen here, along with numerous posts - this is mainly to assist in getting attention on the topic:
   Historic chats no longer accessible after update - #3 by jdubb75
2. Auto-Run/Yolo Mode not working (added in case the initial users concern is swept under the rug, only solution for some users has been clearing the allowList but this is not a fix for all users):
   Auto-run doesn’t work - #10 by yaireo
3. Ongoing complaints about version 0.50+ that are all over the place. A lot of users are experiencing severe slowness, model performance deterioration, oddities within reasoning chains, and rules being fully ignored by models (I am working on a temporary RuleMCP to see if this can improve the rules ignoring issue for some users and will provide it if it seems beneficial):
   Cursor 0.50.5 is a mess for me - #4 by deanrie
   My Experience with CURSOR – Not Just Slow, Practically Unusable Lately
4. Many users complaining about Gemini 2.5 Pro becoming completely unusable:
   Gemini 2.5 pro 05-06: strange hallucination issue
5. Reply to contribute more to the list.

[motic]

I am having all the same issues as the posts you mentioned.

[SergioKingOne]

Slow requests are waaaaay slower than usual.

[jdubb75]

whole heartedly agree, its made agent mode nearly completely unusable.

[hdcrs]

It has been extremely dumb the last few days, used to be awesome and helpful, now it has zero capability of having any context about the project I am working on. Used to be able to scan millions of code lines, now almost zero.

[jdubb75]

are you using any specific model or are you in auto selection mode?

[liquefy]

I can confirm, I made many posts all are ok except the only one thread commenting on significant and highly visible downgrade of o3. post is literally not findable on the forum and gets no views at all

[jdubb75]

Could be the title coming off a little bit aggressively, but all feedback is supposed to be giving cursor guidance on how to provide a good product and maintain their customer base. Can you provide me with the link?

It does always help to use ctrl+shift+p and type capture and send debugging data - that way cursor can get more information to work with. The post does lack quite a bit of detail but its an understandable reason to cause frustration. I do agree, o3 did degrade though.

[T1000]

As a feedback on this post directly:

• Bug reports template in the forum contains specific steps and questions, majority doesnt bother to fill that in. (usually I ask fellow users what they tried)
• Docu pages have a clear list of troubleshooting steps and how to report bugs, majority again doesnt bother to read those or provide info what they tried. (also here I as an user ask others what they tried, frequent no reply)
• Agree that forum should be restructured to better address issues and other needs of Cursor users and the community.

How can the issues you listed be resolved?

There is a huge variety of posts in the forum and many factors affecting responses:

• Lots of ‘historic chat’ posts were addressed and many solved.
• Lots of ‘auto-run/yolo mode’ issues were adressed and many solved.
• Every Cursor version has had complaints for different reasons by different users. Some prefer older versions, others cant wait for new features or updates.
• There are posts for each of the models saying its not working, while others post how great they find the same models.
• In last few months the community was focused on negatives instead of looking for solutions, many expressed how toxic posts/comments were and stopped reading. Several community members who used to help others stopped because of such issues. I took a break for several weeks as I didtn’t feel comfortable how just negative the community got, but it feels much better now.
• Often people add their comments that things are not working on old bug reports that were solved for different Cursor versions and notifying all the thread users of an update to the thread. Maybe a different approach would be better.

For fairnes my reply also boosts visibility of this post.
I am for a healthy and engaging community with hopefully less issues for everyone.
Note that many users are here for positive interactions, helping each other, learning from others. Some posts may get overlooked by forum members because of rants which is natural, not a conspiracy.

[jdubb75]

I agree with you 100%. My main point is that there is potential to systemize or more efficiently handle the issues.

I too try to encourage members to use the capture and debug data feature, while giving more detail about their issues.

The more significant and frequent issues that I do see, there are no pinned posts or continual guidance/trouble shooting that reduces the amount of friction and posting that occurs from users still experiencing those problems - unless I’ve blatantly overlooked them and in that case that’s on me lol.

You are right, a different approach could definitely be better. That is what I’m aiming to get some movement going toward, thank you for the contribution.

[liquefy]

I understand and agree, however I assume it was automatically flagged as it did not even appear on the board, in my opinion filtering for words like ‘useless’ (or whatever caused this) in order to hide negative posts is not the right way to go.
anyway, here is the link O3 is utterly useless

[jdubb75]

I added your post to the list, but yes I agree ‘useless’ shouldn’t come off as too, too strong to be filtered out like that. Posts staying up can perpetuate the negative toxicity T1000 was referring to. If it was automatically flagged, recreating with less strong verbiage could help in the future. Nonetheless, hopefully we can advocate together to continue engaging to have a great product to use for the future.

[T1000]

The forum needs more specific topics from installation, windows/mac/linux versions.
Pinning posts on hundreds of topics wont be helpful for most people who do not have those specific issues. The causes of issues are often so different and not always related to the error shown or issues experienced. Sincerely there are cases ranging from router issue, filesystem read only, no backups,… to various bugs in OS, Cursor, VSCode related problems etc.

But please reconsider promoting this thread on many posts that are not related to your main actual complaint or purpose here. Highlighting this thread does not help those users, its far more likely that they also just get upset that their issue isnt directly addressed while this post does not address them. This triggers further negative comments which results in an avalanche.

The community needs help votes sound good but we as users cant promise other users it will get what they want this way.

[jdubb75]

I agree, there are tons of system-specific bugs that are tough to pin down and consolidate into one post—totally get why that’s a challenge.

I’m also pointing out that there’s no clear pinned post guiding users to send debug data or troubleshoot issues, which makes it harder for folks to report problems effectively when they hit the forum. Even a pinned post that says ‘Reporting a bug? Read this before you post!’ or a pop up during writing into the bug report category that directs them to the pinned post as a backup.

Beyond those user-specific bugs, I’m highlighting platform-wide issues like model performance dropping, YOLO mode not functioning, chat history problems, etc. These seem more like systemic challenges Cursor could address directly, rather than just bugs. My polls show a lot of users are seeing these too (e.g., 100% saying they’re facing similar issues and would like a better system).

Any thoughts on how we could make bug reporting clearer or get a pinned guide up? Would love your take to help the community tackle this!

[T1000]

Sure Cursor team could pin how to report bugs. But my experience here tells me its maybe not the right approach, though I understand why you are asking for this. Given most bug reports without any details its the logical conclusion.

If I recall it right it was also pinned at some time. Also I think that the Bug Report template was updated at least one time since I’m member.

Perhaps that template needs a bit adjustments? For some it may be hard to know what to do, for others it may seem like a Cursor bug but that excludes looking for causes.

For comparison, for me and others I interact here in the forum most of the issues listed here do not occur:

• Auto-run/Yolo works (may need reset if it doesnt)
• Update to 0.50.x created so far most stable version
• Model performance varies but hasnt become unusable
• Chat history never lost

This does not mean the other users dont have those issues, just that they need to be looked at with more information.

[jdubb75]

That would be great honestly, pinning a bug report template would likely declutter and improve some user experiences on here. I believe when you go to make a post there is a template only if you select bug reports as the category, so some users miss this when they post in discussion or other categories. Even allowing users to post their reports in that thread if they fit a certain description or bug type, and mods update them as they get solved. More users might come across a fix for their bug easier that way. Even better categorizing fixes to bugs that are confirmed. Users would need to ensure they’re being prompt as well in the process. If the bug report template isn’t really used then scrub the post.

Any reason you feel its not the right approach? Ensuring the forum gets far less bug reports that lack adequate detail* does seem like a step in the right direction; outside of a way to address the systemic issues I mentioned as they arise and are handled.

Chat history never lost (just cherrypicking the reference), but for some users it is during updates or if they downgrade versions. Under one pinned common issues post though, users could learn methods of preventing catastrophic loss like this more smoothly. For example, using SpecStory extension and backing up chat history.

[T1000]

I have asked on many bug reports what the user tried before, if no info was provided and I usually included links to docu on how to troubleshoot, report bugs etc.

In many cases there was no reply, in others they said they tried everything already .

The bug report still counts but how to get the info needed to triage well known causes that are solvable vs new bug that the Cursor devs have to fix. And this in time before others start commenting (without detailed info) that they have same issue even though the causes may in many cases be different.

Im not saying that there are no systemic issues, but if they were really that many and frequent there would be less positive reports as well, which isnt happening.

Yes chat history gets lost for some users, lets say Cursor team pins that issue with majority of solutions, how would that work for other hundred of issues that others face? Also how to avoid the other less helpful drive-by comments?

Often this is not even an issue of skill or experience level. Experienced users and programmers have also moments with issues or mistakes myself included.

There are cases where bugs occur, but I feel its happening less and not for the same cases which the Cursor team fixed or prevented in their updates.

I proposed a few times more Guide articles and will suggest them again. This should help those starting fresh with Cursor to avoid common mistakes or pitfalls. Like you mentioned SpecStory, but also backups, using Git and commits,…