Subject: Issues with Slow Requests

Hello,
I am a Cursor AI Pro user, but after my premium subscription expired, I noticed a significant slowdown in slow requests day by day. Currently, even when I ask something as simple as the weather, it takes at least 5 minutes to process a slow request. Is there a solution to this issue?

I understand that premium requests are prioritized and faster, but slow requests should still work efficiently. Of course, I don’t expect them to be as fast as premium requests, but waiting 5 to 10 minutes for a single question doesn’t seem normal.

I am looking forward to assistance from the support team.

6 Likes

The only solution they offer is to buy more packs of premium requests, it seems they do not intend to improve slow queue times in any way.

See this post: Slow Pool Information

Yes, there is a serious slowdown when you use the system immediately after your package runs out. but after a few hours this slowdown goes away and returns to normal. there is a warning text that says that it only responds slowly, but do not pay attention to an unnecessary text. my limit is over, but I use it again as if I have premium, sometimes, as I said, it responds slowly again, but I do not wait for it, I turn it off, I turn it on after a few hours and I continue normally.

It seems that way, but if they don’t fix this, they’ll experience an incredible loss of users next month. Acting like 500 premium requests are such a great deal feels absurd to me. Those 500 requests are gone in 3–4 days, and the fact that slow requests take 5 to 10 minutes is almost comedic. I came here thinking I could write faster code, but waiting 5–10 minutes for every fix feels even more ridiculous. I hope they resolve this issue.

2 Likes

That hasn’t been the case for me. The incredible slowness has been ongoing for a week now. My internet is good—fiber connection—so it can’t be an issue on my end. It’s unbelievable that even when I ask a simple question, it takes 5–10 minutes. I contacted support, and for the past three days, they’ve been saying they’ll connect me with someone, but no one has emailed me yet.

ah that’s interesting, hope your problem gets fixed man

türküz la niye yabancı dilde yazıyoruz

Same here… Tho my plan is on business but the slow requests are unbearable

ortama ayak uyduruyoruz lo

1 Like

Absolutely, it’s still insufficient. The slow requests also need to speed up; otherwise, it won’t be adequate in any way.

Just use VS Code + Cline + DeepSeek R1 (using API Key) until you get fast requests or DS R1 API Key usage is enabled with Cursor.

isn’t it expensive?

The ai is so slow I can google and find my answer faster than the dumb ai. It literally takes over 1 minute to generate a simple answer. I hate this if this continues I’m just gonna switch to something else. I don’t have time to sit and wait FOR 1 or 2 MINUTESSSS. For an answer. It’s so slow like a snail.

At this point I’d rather pay for copilot. at least it answers in an instant. Man imagine paying for an ai that you have to wait 1 minute for.

2 Likes

It’s cheaper than running Cursor AI with Sonnet (with either option, API Key or Cursor directly).

And DeepSeek does not run in agent mode on Cursor AI.

So then when you have your fast requests back, switch back to Cursor.

BTW my slow requests are fast rn.

No this is unacceptable. 3 whole minutes for a simple answer. I actually timed it this time.

Paying for more requests just isn’t the answer.

There must be a better solution out there somewhere. I run out of 500 requests so fast. And idk 1000 either isn’t enough. like $40 a month. That’s a lot.

I don’t understand why they can’t increase and make it faster when we literally are paying for it.

Or maybe I just don’t get it. I know ai is expensive so idk

2 Likes

It’s extremely slow, definitely. Sometimes it even takes 5 minutes to answer simple questions.

Yeah. I go through 20 bucks of 500 requests EASILY in 2-3 days. I’m giving them by the end of this month, but yes, I’m already looking at other options. This expense makes AI much less appealing.

EDITED:

Tried R1 as it allegedly should have faster times. Slow requests are as slow as they are on Sonnet, so claim it should be faster is a fabrication.

My trick is to use GH Copilot for all my chat request and keep the cursor 500 fast query for the composer. That I cut down my usage by a lot. Copilot doesn’t seems to have any limit (at least I haven’t hit any yet) on 4o and Sonnet 3.5.

Wow. I guess you lot must use like 2-3K requests a month or something. I went over my 500 requests maybe a week ago? I’m halfway through my subscription and I don’t wait more than 20-30 secs for a response.
I guess you let the AI do everything and/or aren’t programmers?

That’s not the point here, Mr. Voarsh!