Introducing The X Subscription

How do I buy the old Pro version? It was still purchasable just yesterday.

My plan is dead today. I could not use even having $100 credits on it. Grok Bot said Your current plan doesn’t include Grok Bot. I been paying ultra for long time ?

In addition to my previous post, I have a couple of other questions:

  1. Why do we have to use the weekly mechanism? Some users might use it heavily across just a few days rather than having their usage spread evenly throughout the whole month, so a weekly mechanism does not feel very user-friendly.
  2. A follow-up question derived from this mechanism: if a user runs out of their weekly quota but still wants to keep using Grok-Bot or the gork/composer model in Cursor, does that mean they have to use pay-as-you-go quota(Does it charge as expensive as the rawAPI price,even for grok and composer model?)? If a user disables that option, would they have no way to continue using it? Am I right?
  3. After switching to The X Subscription, if Grok Bot uses a third-party model (e.g. a Claude-based agent) internally, is that billed from the weekly pool or from top-ups at API price? And does Grok Bot ever route to third-party models on its own, or only Grok models by default?

Regarding your post, frankly speaking, I agree with most of it. But look at it another way: after trying Grok-Bot for the past several days, I am really attracted by how smooth the whole workflow is.

Even though I am a heavy Cursor user, just as you are, afterwards I might totally switch to Grok-Bot, even for deep, considerate thought on a specific project, because we can do so in Grok-Bot App.

I’m an old subscriber of the older Ultra subscription ($200 a month, with a small discount since I pay annually).

A few days ago, I tried using the Grok-Bot app, and I’m very excited about it. So, regarding the feedback for Cursor or the X team: if they could make the calculation lighter so users noneed to care too much about different models across different app(cursor/GrokBot), it would make the new X subscription tiers much easier to accept.

For example, if there were a subscription tier that includes not only the built-in Grok /composer model but also some extra models(Just like you said), and ensures that whether you’re using Auto in Cursor or using Grok-Bot, it automatically routes to the best model without users having to worry about it, that would be a great direction for the new subscription. It would give at least some new users an easier start with less mental burden.

@Colin Do you know how data sharing for model training and service improvement will work in the new setup?

Brief context:

  • I’m, yet another, Ultra plan user. Cursor is my primary IDE.
  • I also use it for consulting work, so I need to opt out of model training.
  • I tried Grok Bot, couldn’t find settings to control data sharing, and that’s a blocker currently.
  • I was looking into buying an X subscription today so I am generally in favor of this new setup.

Moving to a unified X subscription, I wonder what policy applies and if we can continue to opt-out of model training.

I’m also happy to look at commercial subscriptions if they exist (never checked). I just need predictable cost and clear opt-outs around data sharing or I have a hard stop (which would suck).

Combining Cursor, Grok, and X sounds convenient, but I think keeping third-party model usage included in a plan is important. Model flexibility is one of Cursor’s biggest strengths.

In Cursor, Privacy Mode works the same after switching. If your account is on Privacy Mode (no training), it stays that way when you move to The X Subscription, and your code isn’t used for training.

Training is one setting shared across Cursor and Grok. When you combine accounts, the stricter of your two settings is applied to both, so a no-training choice in Cursor carries over to your Grok account. If you also use the Grok apps, it’s worth opening their data controls after you switch to confirm model improvement is off there too.

There isn’t a separate data-sharing switch inside the Grok Bot app today. Grok Bot follows the Privacy Mode setting on your Cursor account. With Privacy Mode on, your data isn’t used for training and model providers don’t retain it.

2x is marketing speak for we’re removing your 3rd party model bucket and now you get the same allowance but only on Grok.

3rd party models are billed per usage now and you have to pay for them, you pay at API list prices, whatever that means. Probably a quiet way of sunsetting 3rd party LLM usage.

I actually upgraded my cursor plan from $20 to $60 because it will not be available soon IMO, I locked it in for a year as not to be forced to migrate to the new X plan.

The idea of being forced onto a social media platform is insulting.

The transition to the new X Subscription seems to have a serious bug with edge cases. I had a scheduled plan downgrade active from Pro+ to the $20 tier right before the rollout. Instead of a smooth handoff, the system dropped my account straight to Free on Oct 11, permanently wiping out my legacy discount eligibility. Frontline support says they can’t manually override it on the backend. Hoping someone on the team can take a look. @cursor_ai

I feel the exact opposite :joy: I love the way Claude Code breaks it up into session / weekly limits. Even if I go a little nuts and blow through usage, I never have to wait very long for more.

But with Cursor, I’ve had the experience of using all my capacity in the first week and then not being able to use it for 3+ weeks until it resets.

Feature request for product/service

Billing

Describe the request

The weekly reset on included usage is restrictive and creates friction. I pay for a monthly plan, yet I’m forced to perform inside a seven-day window. That doesn’t fit how everyone works.

Projects don’t all move at the same pace — some ship in bursts, others run long tasks across weeks. Why should we have to plan around that? On top of that, it wasn’t part of the deal when I signed up. The marketing talked about at least double the usage of the old Pro plan — but since it’s split weekly, each week you actually get roughly half of what Pro gave you for a whole month.

That’s playing with words. From a user’s perspective, it feels more like cost control for the company than something that helps. Maybe for others it’s a way to spread consumption evenly — I don’t know. But I’d like to hear opinions. We use Cursor because it’s easy.

Please don’t add complexity. If you do, let the user choose whether they want it. Has Cursor addressed this officially? Is anyone else finding it annoying?

Operating System (if it applies)

Windows 10/11

Good post. I got the exact take on Cursor after the X acquisition. I would just add that for me and my partner Cursor won also on speed - using it for fast tweaks and rapid back-and-forth was unreal at the time. But that thing now… “Gruksor” is twice as slow as Claude 5.5 High, and twice as stupid. We’ve been Ultra subscribers for a year now, and this month we both cancelled.

Colin confirmed the 2x usage is spread across weeks, so each week is roughly half of what old Pro gave you for a full month. I get the idea of protecting against one day draining the month, but for a lot of projects — long tasks, sprints, context that builds up — it’s too tight.

My suggestion is simple: let users choose. Weekly reset by default if you want it, but add a monthly reset option for people whose work rhythm doesn’t fit a seven-day window. No new complexity, just a toggle. Those who like the spreading keep it, everyone else gets back the flexibility they had before. Has this been considered internally?

我是取消订阅以后在使用的时候发现无法订阅之前的pro版本了,且合并以后自动为我创建了Grok账户且我不知道账户,而且我电脑已经无法选择购买30美金给服务了,因为页面一直是无显示。

You have built an amazing product, like Cursor, Grok Bot. However, observing community feedback over the past few days, it seems that many users are afraid to use the product due to the reduced token usage after switching to the new plans, leading them to churn to other platforms.

For me, as a user on the $60 Pro Plus plan, the $30 plan is completely insufficient, while the $100 plan is too expensive. The above concerns are regarding the new pricing structure, which is why I am currently staying on my old plan and afraid to switch.

Could you significantly increase the usage limits for the Grok and Composer models in the new plans? Or perhaps provide higher usage for Grok Bot quotas than before?

Describe the request

Hi Cursor Team,

I’m curious whether there are any plans to introduce occasional surprise benefits for subscribers on the new subscription plans.

For example:

Bonus usage credits from time to time

Occasional free quota resets

Special rewards for loyal subscribers

Limited-time promotional benefits

These kinds of perks could make the subscription experience more rewarding, especially for long-term paying users.

Is anything like this being considered for the future?

I’d love to hear your thoughts. Thanks!

Switched from Cursor Pro to Premium. I mainly use Grok Bot + Cursor (occasionally Grok apps).

After the change, everything shares one weekly pool, so usage feels like it burns much faster than on Pro, where things felt more separate. The “at least 2x coding usage” claim doesn’t match the experience when Grok Bot and Cursor both draw from the same pool — same issue others have raised.

Would appreciate a clearer per-product usage breakdown, and whether the weekly pool is really sized for people who use both heavily.

Is anyone from Cursor actually going to provide any updates / feedback about what users are thinking about this new subscription model? It seems like quite a few people are upset by it. Myself included.

I just have a question, @Colin : In Cursor what is going to happen with Composer model? Will we have Composer 3.0 or it will be abandoned because of Grok models?