I’m interested in your new paid project but have a few significant concerns regarding the description of ‘unlimited usage with rate limits’.
Firstly, I believe this description can be misleading. Presenting ‘unlimited usage’ alongside ‘rate limits’ is inherently confusing for users, especially those new to Cursor. It raises questions about whether the usage is truly unlimited or if there’s an underlying, unstated limit.
Secondly, there’s a lack of detailed explanation regarding the conditions that trigger these ‘rate limits.’ For instance, is it tied to a specific number of requests (like the ‘500’ mentioned for the Pro plan), or is it solely based on using models like Claude 4 or the o3-pro feature?
The absence of transparent information on these crucial aspects, such as what exactly constitutes a ‘rate limit’ and the specific triggers, creates considerable doubt for consumers considering a subscription. Could you please provide more specific details on these terms and conditions, and clarify where this information is readily accessible? This transparency is vital for informed decision-making and builds trust.
Yes. From my take here is a third party perspective, as someone currently loving the tool tongue in cheek.
They do not:
Communicate model versions correctly - I have caught substring replacement that trims the version of Google’s Gemini 2.5 - from “Gemini 2.5 Research Pro M_DD etc” to “Google Gemini Pro 2.5” or something similar.
They do not communicate and cause 1 click either to the settings menu or the website to view rate usage.
They do not attempt to manage or translate usage rates.
They do have some sub implementation that affects this, that has dramatically IMPROVED.
I continue to challenge them on being a better model steward, to differentiate themselves from a VS code fork with API calls we could hook up using their tool to ditch their tool.
It’s the sad reality of things, but I don’t want you to be misled, when I usage base use - it is cost insensitive. They don’t really stress this in a way that most humans - if you truly cared about them - need to understand this - especially positioning as a tool that can really create a programmer out of someone that is at day 0 in so many ways. I feel this to be deliberate but that is solely my opinion. I have hounded them about similar issues as the one you raise, for months now. They push other updates, but not these simple ones. Still it is the best tool. But this is the reality - from someone who lives in this tool 12 hours a day. Using cursor is still at your own risk, given I feel no good will in implicity UI communication from an IDE AI company that is primarily marketing themselves for that exact use case. It presents an odd paradigm I’m not sure how to feel about anymore.
The core issue here pertains to the financial and consumption-related aspects of this plan. If there are concerns regarding payment, consumers absolutely have the right to question them. A failure to provide transparent information or detailed explanations about the plan’s contents constitutes a deceptive practice towards consumers.
While I don’t wish to sound overly dramatic, as a user, I am entitled to understand the relevant details. Specifically, I need to grasp how the current practices and the plan are being executed, and what implications these have for both current and potential users.
No I’m not discouraging you - be aware - I’m encouraging you. You are 100% within your bounds.
Communication in a pseudo english auto coder that looks like VS Code, has user expectation, and along with that, really shuffled API usage communication, which I can provide you proof I have continually suggested, leaves me no other option than to question it. At what point is the pig a wolf? At what point does it matter? I intend to be blunt here, the seemingly 40th time I articulate this - it’s not hard.
If someone was selling me snake oil, and didn’t understand snake behavior enough to tell me how the oil flows, that’s an issue. But it’s a larger issue when it isn’t oil. It isn’t a person in front of you. It’s implicit and automatic, and anti-disclaimered by classifying a very easy to use MAX mode (nice name, really?), not communicatiung why 04 mini does 50 tool calls when I am out of MAX mode as cost insensitive, but SEEMINGLY only needs 1 or 2 when MAX usage mode is enabled. The more I think about it, the more I classify it as totally inappropriate again. I see saw so hard. This is important stuff though. What is a person supposed to infer here?
Objectively I think I get to say here, almost a year into some of this - one of the following is true:
I believe that Cursor’s complete lack of detailed explanation regarding their subscription plan content, how it’s executed, the timing of triggers, how conditions are met, and what limitations follow achieving certain milestones – among other factors – indicates that their company is fundamentally built upon “misleading consumer practices.” Every time a new feature is introduced, it leaves consumers feeling confused and uncertain. Furthermore, when questions are raised, it’s typically not their development team providing explanations, but rather community managers, customer service personnel, or individuals not from the core management tier. I find this approach to business operations to be quite poor.
In the United States, where Cursor is based, consumer protection laws, particularly those enforced by the FTC, mandate that businesses must not engage in unfair or deceptive acts or practices. This includes providing clear and accurate information about product features and limitations to consumers, especially for subscription services where expectations around usage are critical. The lack of transparent details on rate limit triggers and definitions could therefore be considered a violation of these principles.
Couldn’t have said it better myself. That’s the implication I’m led to. This is an IDE company, training or offering models on high level web dev and app dev patterns, in well common understood mechanisms:
Languages and Frameworks
Terminal Tool Calls
Implicit UI Communication (an IDE would be considered Expert difficulty, imo)
Complicated API stewardship and Cloud Hosting or
Prompt Engineering
Embeddings Infusion
Agentic Handling with deterministic if/then IO behavior, which is as old as computers
etc
At the cornerstone of all this, yes, we do get to demand written explanations for all of these things. Improvements aside. What’s going on yall?
Everything they do, they should disclose if it affects the API spigot. Their prompting, their caching, their handling of context, disclosing what they do above and beyond the browser other than terminal commands they don’t let me alter.
My rules are ignored. They probably don’t know why - but you just cannot continue to slap embedded content over embedded content, when some are doing internal and some external things, and some are liberties taken for the consumer, while others, much easier, clearly, are not. I mean I am not a web dev, I did make the consideration today about my users and API usage and what my responsibility was in that, legal or not. Just saying. Development is such, even in AI, that I have enough time to not be able to avoid considering the implications of my design. It is literally the ENTIRE point of web dev. So regardless, pig or wolf, it needs fixed because it is the EASIEST of all of this.
Prompt Suggestion: “How can I create a quick meter that tracks user API usage? Currently, it seems a bit odd the most crucial thing to my user - its wallet and resource usage - is behind a click. It also requires a user to click to our website to infer indirectly with model information what their API usage may be or what cost insensitive means, or if perhaps, system interrupts are corrupting their tool calls causing massive churn? Not sure it is, but can we look into it? I notice plenty of empty space near the chat, both above and below - we have options.”
If the policy has been implemented immediately without providing any explanation regarding its impact on the current paid plan, nor detailing the actual functionalities of ‘Burst rate limited’ and ‘Local rate limited,’ does this already infringe upon consumer rights and violate consumer privacy laws?
This thread looks like shadow-banned (unless something wrong is going on with the quotation marks). 4 days, only see the thread you brought up link to your question.
I would recommend commenting on less censored platforms like X or chatGPTCoding subreddit (cursor subreddit is “positively”, as local mod called these shadow-bans, censored too).
Thank you for replying this to me. I somehow realized that they avoiding my question since the plan announced then. I knew this will coming but never thought about they will ignoring customer’s thought and banning them for all.
