I have been using Cursor since 2023. Cursor has been part of my development workflow for almost three years now. During that time, I have also used Codex, Claude Code, Amp, and OpenCode, but I kept coming back to Cursor because I genuinely felt that the product had a level of polish and attention to detail that others often lacked. I have recommended Cursor to many people for exactly that reason.
But looking back at the history of these pricing and plan changes, I unfortunately have to say this very directly: Cursor has repeatedly failed to maintain user trust.
From the original “500 requests” plan that many users understood as a clear and stable subscription model, to later changes that felt almost forced onto existing users, and now this Max Mode requirement for legacy individual plans, the pattern is becoming very hard to ignore. Maybe the product team and the marketing/communication team are completely different teams — which would at least explain the disconnect — but from the user side, the experience is simply inconsistent and frustrating.
The current change is especially confusing. If the concern is model cost, a much more reasonable approach would be to assign different request costs to different models. For example, a more expensive frontier model could consume 5 or 10 requests instead of 1. That would be understandable. Users can accept that different models have different costs.
But completely blocking traditional request-based subscribers from using frontier models unless they switch to Max Mode feels like a disguised way of pushing legacy users out.
What makes this even harder to understand is the logic behind it. If users heavily use a very expensive model like Opus 4.6, that can also create serious model cost pressure for Cursor. So why is that acceptable under the current structure, while other frontier models are completely locked behind Max Mode?
And if another newer frontier model has input/output pricing that is lower than or comparable to Opus 4.6, why not make it available under a similar request-based pricing rule? Why force Max Mode for those models as well?
GLM 5.2 is a good example. Its pricing appears to be lower, yet it still requires Max Mode. Have you actually used Max Mode as a normal user? In practice, once Max Mode is enabled, almost any model can consume the quota extremely quickly. Sometimes it feels like one serious request can burn through a huge part of the allowance.
So I genuinely do not understand the thinking here.
The issue is not that users expect unlimited access to every expensive model forever. The issue is that Cursor keeps changing the meaning of plans that users already paid for, while presenting the changes as technical corrections or pricing necessities. That damages trust.
Cursor is still a strong product. That is exactly why this is so disappointing. But a strong product does not excuse weak communication or repeated changes that make loyal users feel pushed aside.