Hey Cursor Team & Community,
First off, Cursor is an incredibly powerful tool when it works as expected. It’s designed for deep work and efficient coding, which is exactly what I need as someone who pushes systems hard and relies on AI for complex tasks.
However, I need to raise a critical issue regarding the new Pro plan ($20/month) and the severe rate limits being applied to Claude 4 Sonnet (specifically the version without Thinking enabled). Adding to the frustration, it seems the old pricing plan has also seen a dramatic, sudden price increase for the Thinking model.
My Experience on the New Pro Plan (Today):
I’m on the new Pro plan, expecting a certain level of usability for my workflow. Today, I was using Claude 4 Sonnet (standard, non-Thinking). My usage was minimal – maybe around 5 to 10 requests within a relatively short period (less than an hour, certainly not hours of continuous hammering).
Suddenly, I hit a rate limit and received the message:
"You’ve hit the rate limit for your Pro plan. Upgrade to Ultra for near unlimited use, set a Spend Limit for overages, or switch to auto"
The Problem - A Dual Crisis:
- Extremely Low Limit on New Pro Plan: Being rate-limited after only 5-10 requests of Claude 4 Sonnet (non-Thinking) on a paid $20/month plan is unacceptable for any serious user. This isn’t “Pro” usage; it barely covers basic debugging or a short coding session. It completely breaks workflow and focus.
- Dramatic Price Hike on Old Plan (No Escape Hatch): As discussed in other threads, the price for Claude 4 Sonnet Thinking on the old/legacy plan has reportedly jumped from around $0.08 per use to $0.80+ per use (a 10x increase, potentially even higher based on some reports like mine seeing over $1.60 per request before hitting the limit). This change seems sudden and lacks clear prior communication. This means that for users who rely on the Sonnet Thinking model, switching back to the old plan to escape the new plan’s rate limits is not a viable solution – it just becomes prohibitively expensive instead.
- Lack of Transparency (Both Plans): The absolute lack of clear information on what triggers rate limits on the new plan and how long cooldowns last is still a major issue. Similarly, the sudden, unexplained price change on the old plan without clear justification or advance notice is unprofessional. We need to understand the system’s mechanics and pricing structure to use it effectively and predictably.
- Value Proposition Decimated: Between severe, opaque rate limits on the new plan and a sudden, massive price increase on the old plan for the Thinking model, the value proposition for users who rely on Claude 4 Sonnet (especially Thinking) is severely diminished across both available options. It feels like users are being pushed towards the $200 Ultra plan or unpredictable, expensive usage-based billing, neither of which fits everyone’s needs or budget.
Why This Matters for “Tryhards” and Power Users:
Users who leverage Cursor for deep, complex problem-solving – pushing the AI to its limits to build innovative solutions and tackle difficult problems – require reliable, consistent access to powerful models like Claude 4 Sonnet. Being throttled after a handful of requests on the “Pro” tier, or facing a 10x+ price increase on the legacy plan, feels like being told to stop innovating just when things get interesting and productive. It disrupts the very workflow Cursor is supposed to enable.
What We Need:
- Transparency: Clear, documented rate limits for each model on the New Pro plan (e.g., X requests per Y minutes/hours) and predictable cooldown durations.
- Review of New Pro Plan Limits: If hitting the limit after ~10 Sonnet requests is the intended behavior, then the Pro plan description needs to be much clearer about these severe restrictions, or the limits need to be re-evaluated to be actually “Pro”-level usable for sustained coding sessions.
- Clarity and Stability on Old Plan Pricing: A clear explanation for the sudden and massive price increase for Claude 4 Sonnet Thinking on the legacy plan, and ideally, a reconsideration of this pricing to make it reasonably affordable for continued use for those who prefer that model/plan structure.
- Viable Options: Users need either a usable “unlimited” tier (with generous, transparent limits) or predictable, affordable usage-based pricing that doesn’t suddenly jump 10x+. The current situation offers neither for those who rely on Claude 4 Sonnet.
I know I’m not alone in this (seeing the discussions in other threads). This isn’t just a minor bug; it’s impacting core usability and trust for many of us.
Cursor Team, can we please get some concrete information, transparency, and a commitment to address these issues across both the new Pro plan’s rate limits and the old plan’s price hike?
To other users experiencing either the severe rate limits on the new Pro plan or the sudden price increase on the old plan for Sonnet Thinking: Please share your specific experience below. The more visibility this gets, the better chance we have of getting a usable solution.
Thanks for listening.
A frustrated Cursor Pro user who just wants to build things without hitting invisible walls or facing sudden, unaffordable costs.