If you don’t mine, I’d be interested in your experience with Kiro. Why do you say Kiro is already better? What does 1000 “credits” get you with Kiro in terms of requests? How would you compare it in terms of the number of requests you can get with Cursor’s $20 plan? What models are available?
Just a month ago or so Kiro didn’t use credits (used to be vibe requests), so they are still adjusting and may like Cursor adjust their pricing plan to survive. Going from explicit requests to something more vague and variable as “credits” is telling of the direction they are headed, like Cursor. However, it would make sense that Kiro is trying to give more for your buck in attempt to attract customers in the short term.
Regarding the latter in your post, it’s exactly what Cursor did. Kiro simply returns more for value now, if it changes, so be it. Perhaps AI was never really meant to be affordable.
I am a professional but I guess that’s why I am not as cost sensitive as you. I see the current pricing (including on-demand/overage cost) as a positive ROI. I don’t even care about limits happy to pay more. I notice that most of the folks that have issue are cost sensitive as they are hobbyist with perhaps limited coding experience or they are in developing nations where salaries cannot justify the expense. I wish it accessible by all but again the ROI should justify current cost.
I just get tired of reading the same “its too expensive, there are better alternative” complaints. If there is a better alternative that is cheaper go use that and post in their forum. Its entirely valid to share an opinion on pricing, bugs, features, what you don’t like but these discussions never come across constructively.
I’ve been around long enough to know when i’m being ripped off. It’s like paying 10 times more for jeans because there’s some designer logo on it, while there’s better jeans out there. Sadly there are many who fall for marketing and truly believe the designer jeans are better in every aspect. They’d even defend the billion dollar company. But hey, as long as the ROI is good, may ignorance be a bliss.
GPT-5-high requests cost me ~$0.30 per request and save me 15-20 minutes of senior-level work. I get 60 of them per month (~20 hours), which is a ~$2,000 value.
8 months ago, Claude 3.5 requests included with Pro cost me ~$0.04 per request and would do 10-15 minutes of junior-level work. I would get 500 per month, which is a ~$5,000 value. BUT, 80% of my work is senior-level, so the real value to me was closer to $1,000. In order to get more value, I would have to pickup odd junior-level work from the backlog, which was not optimal. There was a reason that work was not prioritized.
3.5 is perfectly capable of doing senior-level work. When 3.7 came out many were still using 3.5. Maybe the plan and tab features make it “senior” for you, but 8 months ago I was working on a SaaS and had great results. So were many more. New pricing = rip off.
Can you provide us how cost effective Kiro is? Number of requests, with what models you can expect to get on their 1000 credits plan? You clearly have determined that Cursor is a rip off by comparing it an alternative (Kiro).
I honestly don’t know the people commenting on my post. Maybe they’re junior developers who find Cursor helpful for web development, or possibly PR folks from Cursor itself, since I’ve seen that same persons actively defending Cursor in many other threads. Most brands do this kind of promotion, so it’s nothing new.
As I mentioned earlier, in terms of quality and understanding prompts, Cursor is definitely good. But providers like Trae offer 600 prompts for just $10 using the same models that Cursor uses. Cursor is essentially a wrapper and is heavily profiting right now.
Mark my words — once other platforms reach Cursor’s current quality (and they’re improving fast), Cursor will have to lower its prices to stay competitive. For now, they’re just trying to collect as much profit as possible.
No one knows each other. Not sure how that proves that the replies are some Cursor PR stunt.
I tried Trae, it was god awful and has no C# extension that worked with its VS Code fork. Yea they are cheap, but its almost a guarantee they are training off your code (if privacy is a concern) since they don’t say they don’t explicitly like Cursor does. Also when I tried Trae, it was super slow and I was in a queue for like 10 mins, but that was a promotional model, but I have never experienced anything like that with Cursor’s promotional models.
The opposite is going to happen. Trae and Kiro are not sustainable, they are the ones trying to gain users from Cursor before they run out of money. Cursor already did that and has a stable market share, so yea, Cursor is now at the point where they are trying to earn a profit. But Trae and Kiro are still doing big promotions and changing their pricing and policies. Eventually they will raise their prices or dial back their quality, that is my assumption. It’s just not sustainable if you look at what the AI model providers are charging themselves. 1+1 cannot equal <2 if these companies want to turn a profit. Promotional deals with providers and VC money will only last so long.
Also Cursor is just overall better for actual programmers from my experience.
JetBrains IDEs weren’t even wrappers when they already cost a decent amount of money annually. You pay not just for API access to providers but also for convenience and QoL features as well as Cursor-specific services like Tab autocomplete.
I can’t say how much they are profiting but it seems like they at least do profit now while in the past they were losing money to gather users. Trae, Kiro and others are heavily subsidized by the providers or even from their own pocket (well, their investors’ to be precise). They are in the phase Cursor was and they either die or become like Cursor making their pricing fair on them.
They are a different product. As long as they have something that sets them apart from competitors (their awesome Tab autocomplete for example) they won’t have to lower their prices as they will be at a different niche.
And if they lose this advantage (for example, if Kiro releases autocomplete that is on par with Cursor’s) – sure then, they will start losing clients fast and will have to change something to stay alive.
It’s a business. If you think they are ripping you off – you can choose a different product. If enough people will agree to you and do the same – they will adapt.
When you see them doubling down on their pricing changes – that means that you are in a minority and a lot of their clients are happily choosing to use their product instead of competitors’. It’s not because you’re wrong, it’s because you crave for a different solution with a different price while they are satisfied with current one.
We all have different use-cases. These products cater to different audiences.
And no, I’m not a PR person, they couldn’t even hire me if they want due to my country of origin lol. I’m just a long time user who is happy that they slowly shift away from the curse of “vibe-coding” and develop their product back to my liking.
Why do y’all have to insist we are junior developers. I am not and cursor is my main IDE. I flip between cursor and vs code with codex. Now one difference is I don’t like to give coding agents a huge chunk of work and expect to implement in one shot. I usually have a pretty good intuition of how I want to build something and I will use the model to build chunks of functionality with fairly specific requirements. I get a ton of value out of Cursor doing this. I do typically use Auto as its gotten fairly good for my requirement so we will see how that will change in terms of cost but I am fully ready and happy to pay for the next tier, it drives that much value for myself.
Or we’re professional senior devs who A) know how to use Cursor efficiently, and B) recognize that AI is actually very expensive to operate, and any Cursor competitor offering much lower prices is taking massive losses to grow their userbase.
You got it completely backwards. If other platforms reach Cursor’s current quality, they will raise their prices so that they can actually start making a profit instead of bleeding money.
This nails it imo. I have built a number of backend systems that touch LLM apis and its not cheap when you start throwing tokens at it. I think sometimes people don’t realize the number of tokens being used in the chat interface and the cost of those tokens. I am sure there are ways to optimize what is being sent but its not easy and I also like to think that its in Cursors best interest to optimize it.
Not sure how that proves that the replies are some Cursor PR stunt.