Frustration with Cursor's UE

Feedback
1

I’ve been an early adopter of Cursor, enthusiastically signing up for the Pro version and even prepaying for a year. Initially, everything was fantastic, but lately, I’ve hit a wall with constant throttling. The delays have been unbearable, with responses taking anywhere from 3 to 10 minutes (if ever). I often find myself having to interrupt and resubmit questions, which is maddening, especially considering…

In an attempt to resolve this, I purchased APIs from both OpenAI and Anthropic, where I’m also a Pro customer, paying $200 a month for ChatGPT and the Pro version of Anthropic. Despite this, Cursor doesn’t allow me to fully utilize these Pro versions through MCP or any other direct integration. Instead, I’m funneled through Cursor’s Pro plan or pushed to buy additional API credits. But the frustration continues: when enabling these forced double-dip APIs in the settings, I’m greeted with a message claiming Pro users don’t need them. This is gaslighting at its finest. If the cursor Pro version worked as advertised, I wouldn’t be forced to rely on additional APIs just to get a basic, usable experience.

Adding insult to injury, now there’s a constant prompt to pay extra for “fast responses.”. Can we say triple dip? What in the actual? I’m already on a Pro plan, which should provide a functional experience without these constant upsells. What am I missing? Does cursor feel good about this trash they are serving?

2

You nailed it. I’m in the exact same boat — long-time user, Pro plan subscriber, and someone who genuinely wanted Cursor to succeed. But lately it feels like the product is collapsing under its own weight. The throttling is unbearable, the latency completely kills any workflow, and the “fast response” upsell is just insulting when we’re already paying premium prices.

I’ve also tried every possible workaround — external APIs, custom setups, you name it — but it’s clear the system is locked down to force users into paying again and again for less performance. The irony is that I want to keep using it; the early days showed real promise. But now, between the constant lag, the mixed messaging, and the paywalls inside paywalls, it’s just not a sustainable tool for professionals anymore.

At this point, it doesn’t feel like an AI coding assistant — it feels like we’re beta-testing a monetization experiment.