The burnout and tone of response by the mods is pretty telling. Unlimited IS a hidden price increase. Going to attempt to get a refund, we got rug pulled.
10 Likes
I think we all know what must be done. My question is, what is our best option?
1 Like
We’re currently facing a situation where we’re being presented with seemingly improved conditions, but in practice, it’s a case of giving with one hand and taking with the other.
Discontinuing services that no longer provide value might serve as a signal to reconsider the approach toward customer relationships.
If we, as clients, accept these changes without question, it may lead to further losses and a continued decline in transparency.
6 Likes
I signed up right before the shift to “unlimited,” and I haven’t used the larger Claude models—just Sonnet (the mid-tier one). Despite that, I’m still regularly getting locked out of all Claude models, which makes it hard to justify the “unlimited” label or know where the limits actually are.
To make things worse, different Claude models behave very differently depending on the task. So using auto-selection isn’t a good fallback—it doesn’t reliably switch to something that works. Instead, I end up with broken workflows and no clarity on what triggered it.
Right now, the official documentation (docs.cursor.com) doesn’t give any real transparency on rate limits or throttling. There’s no clear info on:
- What triggers a lockout (request count? token count? compute time?)
- How long limits last or when they reset
- What “slow mode” means or how it impacts speed and access
- Whether different models have different usage budgets
Instead, we’re left guessing based on vague language like “burst limits refill slowly” or “local limits reset every few hours.” That’s not useful for users trying to understand why the product isn’t working when they need it to.
From a small business perspective, the “just throw more money at it” approach isn’t helpful. If I can’t even see what I’m paying for at the Pro level, upgrading to Ultra feels like gambling—not a smart or sustainable investment. Businesses need predictability and transparency to make decisions, not a black box that sometimes works and sometimes doesn’t.
Legally, this lack of disclosure is questionable. Under U.S. consumer protection law—especially FTC rules under 15 U.S. Code § 45(a)—companies are expected to disclose material limitations on paid services. Rate limits, throttling, and access restrictions clearly fall into that category. The FTC has acted against companies that obscure usage limits or degrade service without warning (e.g. AT&T, TracFone).
At the very least, Cursor should publish:
- Concrete usage limits by plan
- Throttling thresholds and reset timelines
- Model-specific differences, if any
- A way to monitor current usage (even roughly)
Without that, it feels like users are getting penalized for using the product effectively—and left in the dark about how to avoid it.
6 Likes
This simply does not answer a single question that is being asked in this thread.
7 Likes
I’m also using the pro plan. But it’s not max or opus, it’s just the normal variant Claude 4, it’s not even thinking about it and it’s over in 1 day. Then switch to free ChatGPT 4.1, he says. I think this new wage policy is currently a headache for the cursor team. Because there are so many complaints.
4 Likes
这个透明度是很有问题的，完全的黑箱操作，文档模棱两可，介于此我已经在做相应的其他方案了…
This level of transparency is a problem, completely black-box operation, ambiguous documentation, and because of this, I am already working on alternative plans…
2 Likes
I just paid my bill yesterday, and got rate limit in short time, I am so confused and the user experience is getting worse and worse. Ugly eating manners !
Brothers, it’s time to find a replacement. Trae is more transparent, and the price is less than half of Cursor. Although it may not be as powerful as Cursor, based on my usage, it has already achieved 75-80 out of 100 points, and it is still continuously iterating. Let’s say no to black box operations. 
6 Likes
After testing Claude Code extensively, I don’t see any reason to use Cursor’s new $20 pricing plan compared to Claude Code at the same price point.
Claude Code delivers:
- Much better context handling for large codebases
- Predictable usage patterns with transparent 5-hour reset cycles
- Reliable limits - you actually know when your capacity refreshes
- Consistent performance without unexpected rate limiting
I’ve reverted to Cursor’s old pricing system, but if they remove legacy pricing, Cursor becomes absolutely unusable for serious development work.
The new model isn’t just worse value - it’s fundamentally unreliable and breaks development workflows.
The math is absurd: I’m hitting limits after 30-40 minutes of work on the new $20 plan. Cursor then suggests upgrading to Pro+ for $60 to get “3x more capacity.”
So for triple the price, I get… 90 minutes of work before hitting limits again? And I have no idea when these limits reset! 
Meanwhile, Claude Code gives me reliable all-day coding sessions for $20, or their $100 plan for 5x more capacity - both with transparent limits and predictable resets.
11 Likes
My billing cycle reset on 27 June. But yesterday (July 3rd), one week later, I’m already locked out of Claude-4 and 3.5. The pop-up just tells me I’ve “saved $94 this month” (
) and that I’ve hit a rate limit, yet it never explains what limit I crossed. Was it tokens, dollars, requests, or some burst quota? Worse, there’s no warning, dashboard tracker or running tally to show where my usage actually stands.
Without that information it’s impossible to plan. One minute the models work, the next they’re shut off, and I’m left scrambling for alternatives right in the middle of a project.
Immediate need for all Pro plan members:
-
A live usage page that shows tokens or cost in real time, plus when the limit resets.
-
Clear, published numbers for the Pro-plan caps both daily and monthly.
-
A warning when we’re approaching those caps instead of an abrupt lockout.
If anyone has figured out a way to see their actual usage, or if the team has an update on when better visibility will arrive, I’d love to know.
Update [July 4 @12pm]:
15+ hours still unable to use. I have no idea if the block is for the whole month!
7 Likes
I will not renew my subscription if there is no transparent pricing, especially one that doesn’t disrupt my workflow.
1 Like
Is it me or is this post now ‘hidden’ from topic, search?
4 Likes
I am absolutely shocked that this post has been excluded from visibility. This is another serious blow to Cursor’s transparency. The company is behaving like some of the worst dictatorships on the planet when it comes to censorship. Maybe we should meet and organize elsewhere like telegram to discuss alternatives. I don’t think any company should treat its customers the way it does here and the consequence will be that many will cancel their subscriptions.
7 Likes
Someone should post this on Reddit (I don’t use it), so that it can gain more visibility there since they’re trying to block it here. Twitter too.
3 Likes
Twitter works much better, I guess.
2 Likes
Update - 3:15 PM EST July 4th: Rate Limit Finally Lifted After ~26 Hours
Thank You to Everyone
First, I want to thank everyone who contributed to this thread and will continue contributing as this issue remains unresolved. The response has been overwhelming and shows this is clearly a widespread issue affecting many Pro subscribers.
Special thanks to @PabloT @dbblackdiamond @Xernive @t0ny1 @Kirai @Alexandre1 and many others including Developer_Spooky, fusiondev, D.V, Nazruden, NahNah, curtdev, decsters01, Gohrd, DesignGears, pleiadess, wtester, b_l, Liam_Baugh, Genshro, murongtianyu, qomosoloto, rahit, malcsi, datrim and everyone else who shared their experiences, screenshots, usage data, and frustrations.
Your detailed experiences, screenshots, and support have been invaluable in documenting these transparency problems and proving this affects the entire Pro subscriber community, not just individual cases.
Addressing Staff Response
@danperks - Thank you for finally responding, though your answer still doesn’t provide the specific numbers we need:
- “Every few hours” still means nothing - Is it 4 hours? 8 hours? 24 hours? We need exact timing.
- “Greater than your plan allows” - But what IS the plan allowance? How many Claude 4 Sonnet requests can Pro users make?
- Auto model suggestion - Many users have explained why this doesn’t work for their workflows.
@T1000 - I’m disappointed by your dismissive response calling this “conspiracy theories.” Multiple users provided concrete evidence of shadow banning, including @Kirai’s documented experience. This isn’t conspiracy - it’s documented fact that you can verify yourself.
The thread being forwarded “to Cursor Team” after 10+ hours and multiple users confirming the same issues shows the system isn’t working as intended.
Thread Hidden Again
Once again, this thread has been hidden from public view as confirmed by @malcsi and @datrim with screenshots. The AI moderation system clearly needs adjustment if it keeps hiding legitimate customer feedback about service issues.
Please fix the AI that automatically hides posts - this is the third time this professional discussion has been suppressed.
Rate Limit Status Update
Good news: When I opened Cursor today at around 1:15 PM EST, I tested Claude 4 Sonnet and it worked without issues. The rate limit appears to have lifted after approximately 26 hours (from 11:19 AM July 3rd to ~1:15 PM July 4th).
Bad news: I’m now scared to use Cursor normally because I don’t want to hit another surprise rate limit. If these limits last an entire day (as my experience suggests), it makes Cursor unreliable for consistent development work.
This confirms that “every few hours” actually means “approximately 24 hours” - which is completely different from what the documentation states.
This Is Not Just a Forum Issue
The same problems are being widely discussed on Twitter/X. This is clearly a systemic issue affecting the entire Cursor community:
Evidence from Social Media:
- 0ni_x4: “rug pulled??? Making the pricing ‘unlimited’ and then changing it to limited without any warning” - https://x.com/0ni_x4/status/1940885976127283342
- gabriel__xyz: “just 12 days after this, they changed the pricing back WITHOUT notifying anyone” - https://x.com/gabriel__xyz/status/1941011597142765626
- oscarle_x: “$7k yearly subscription. And then you immediately pull the rug on us” - https://x.com/oscarle_x/status/1941077262079537586
- ravishar313: “I just paid for the Pro plan on 1st. I have not used it this much, wtf is this” - https://x.com/ravishar313/status/1941040959770538251
- hieudinh_: “removed all transparency regarding the number of requests or agents you can use” - https://x.com/hieudinh_/status/1940992222021013809
- Navetz: “burned through my subscription in 3 days when normally i make it like 20-25 days” - https://x.com/Navetz/status/1940900671597039733
- umbrella_uni: “Cursor was unlimited before now they want to charge you for each use” - https://x.com/umbrella_uni/status/1941045832444477744
- seconds_0: “Entirely silent change of terms with no communication” - https://x.com/seconds_0/status/1940872503448293892
- iamgingertrash: “Cursor is becoming a horrifyingly greedy company” - https://x.com/iamgingertrash/status/1941162961349095730
- startupseriesio: “Cursor’s recent changes are a masterclass in what not to do!” - https://x.com/startupseriesio/status/1941082832585289825
- btibor91: “‘Unlimited agent requests’ has been removed from the Pro pricing page” - https://x.com/btibor91/status/1941101695859393012
- DodgeHealer: “silently switching the unlimited pro plan and overcharging for 6 sonnet request (10$) is crazy” - https://x.com/DodgeHealer/status/1941099497087816176
- DanielLockyer: “I have no idea what the actual limits are anymore” - https://x.com/DanielLockyer/status/1941067519571419176
These are just a few examples - there are dozens more tweets with hundreds of replies expressing the same frustrations.
What This Proves
- The issue is widespread - affecting users across all platforms
- The documentation is misleading - “every few hours” actually means ~24 hours
- Silent plan changes - “Unlimited” was quietly changed to “Extended limits”
- No transparency - users can’t track usage or predict limits
- Community frustration - users are actively seeking alternatives
Moving Forward
The evidence is overwhelming that this transparency issue affects a significant portion of your user base. Users are:
- Seeking refunds and considering chargebacks
- Switching to alternatives like Claude Code
- Warning others about the pricing changes
- Questioning the company’s trustworthiness
Requests for Cursor Team
- Provide exact reset timing - Replace “every few hours” with specific timeframes
- Show actual usage limits - How many Claude 4 Sonnet requests can Pro users make?
- Fix the AI moderation - Stop auto-hiding legitimate customer feedback
- Unhide this thread whenever the bot incorrectly flags it
- Address the community concerns shown in the Twitter evidence above
The community has provided clear, documented evidence of systemic transparency problems. The response needs to match the scale and urgency of the issue.
This thread should remain visible so other users can contribute their experiences and find solutions to these widespread problems.
17 Likes
Holy hell if this doesn’t get seen by a moderator or someone on the cursor team to explicitly prove that this was a mistake on this “forum bot” that “censors” posts etc. though however ironically the ones that (always matter in importance…)?
I feel absolute sorry for anyone who hasn’t figured out how to cheat the legacy system and the new rates etc. because you’re going to burn $$$$$$$ insanely quick. How can this go unseen and swept under the rug when the community is doing their job on providing issues but when the issues are being blatantly ignored, what else can be done?
This forum seems to be more of a jerk me off fest with bots and fake ai responses than real human being intelligence or logic being used over the counter.
Wouldn’t surprise me if the owner of cursor is off in the Bahamas sipping on a margarita laughing at us.
@T1000 : I respect you at least trying even though you aren’t being compensated for our outrages and time of need etc., real or not real you have done a great job in showing activity and giving us a sliver of hope that someone out there cares for us. You have my respect, thank you.