Just buy Claude Max $200 plan and add to Cursor and reduce your cursor sub down to bare minimum.. Its the only way for real complex work now because Grok just simply IS NOT READY for that level like Opus/Sonnet is. I HATE saying this! Feel like i’m betraying cursor but at the end of the day - they did this to us.
I think that slogan of “subsidising tokens” has been vastly overblown. Probably to increase the perceived value of the service. Personally we weren’t able to even hit 50% of the monthly quota given that Claude took over most of the usage for us. Another argument is that majority of our business hours is being spent on implementing all that software we built earlier. Meaning we’re not using that many tokens like we used to at the beginning. It’s changed to bursts of intense work and then lighter token use as we divert focus from dev to implementation.
Finally, there is a reason they sell the new model with “2x quota” now. Surely that wouldn’t be the case with all these supposed subsidies going on… ?
I ended up doing something similar, downgrading and using the savings to go around Cursor for external model access.
I can’t conceive of why X would do this rug pull, external models are essential and a big part of the value prop with non-minimum plans.
I’ve been on paid subs with Cursor and X since 2023, and this move just cut their revenue from me substantially. The duplicitous nature of the unification flow in falsely claiming plan comparability erodes trust, compounded further by the stonewall responses from support.
I fear you guys are going to destroy what makes Cursor so unique by moving in this direction. I hope you don’t force the change for Cursor users. I know you’re probably caught in a weird spot after the SpaceX takeover, but it’s worth saying as a long time customer.
It would be helpful to see visually what i could be saving if I did switch based on my real time usage. i already pay for X premium, so perhaps for $92, if Grok 4.8-5 ends up being sufficient for my needs, I’d save money over the min $200/mo I’m currently paying. But I have no idea what 15x usage means, and also the loss of third party included pool is a giant question mark.
Can’t help but feel i’ll be paying many hundreds more for the same output.
Dont switch, keep your legacy account. Do not make the same mistake hundreds of us did.
@Deviation - Thanks for the suggestion. I am definitely going this route, or if I want to get away from Cursor completely, VS code with the Cline extension with a Claude sub attached to it.
It’s bad enough that they did this to us, but the worst part is there are no responses at all about any of it. It’s like they don’t even care about their customer base.
If this was clearly notified that it was coming, and we had time to prepare, or at the very least get a prorated refund because the service has changed, then it wouldn’t feel like such a betrayal.
Hey everyone, thanks for all the feedback in this thread, and for your patience while we worked through it.
To be clear: for existing subscribers, switching is opt-in. If you’re on a current Cursor plan, it keeps working exactly as it does today, including its third-party model usage, unless you choose to switch. You can also keep moving between legacy plans (Pro → Ultra, Ultra → Pro+, etc.). There’s no timeline for migrating existing plans, and we’ll give advance notice before anything changes.
From the start, we’ve designed the unification flow to spell out exactly what happens if you unify. Before checkout, you’ll confirm that third-party models aren’t included in the plan price and are billed separately based on usage. Switching is permanent, and legacy plans can’t be repurchased once they end. If third-party models or a monthly usage pool matter most to you, staying on your current plan is a perfectly good choice.
For those who’ve already unified: we’re looking into reports of usage limits running out faster than expected. Some of this appears to come from people trying newer features included in their plan (like Grok Imagine), and some we’re still digging into. Limits that run out quickly aren’t good for anyone, and we want yours to hold up for real work.
If something went wrong with your account specifically (a charge, missing Bots, or a plan that didn’t carry over), email [email protected] with your Cursor account email and the X or Grok account you linked.
This doesn’t help people that were already rug pulled by a bad unification trade-off that was explicitly labeled as comparable.
This gas lighting response further erodes trust.
we’ve designed the unification flow to spell out exactly what happens if you unify. Before checkout, you’ll confirm that third-party models aren’t included
The flow does exactly the opposite, it highlights the plan as comparable, which it is not.
The copy I recall seeing about external models was ambiguous at best. External models weren’t “included” in Pro+ either, having their own rates separate from Grok/Composer.
The end result is beyond contestation, people are downgrading because you’ve rug pulled considerable value. Reconcile.
Yeah colin thats all well and good but youre not reading what we are saying. GROK is NOT good enough for complex work yet! Have you not seen how many of us are cancelling because its destroying our work / repos?
I understand pushing your own models with SpaceX etc… but do it WHEN GROK is UP TO STANDARD please. Not when it is failing on millions of tasks, spitting out slop all round.
@Colin, the wording around 2x Cursor usage and more usage-per-dollar is not something I’m experiencing based on how fast the quota is draining. Cursor plans included separate weekly Grok Bot usage. One would have expected 2x (Cursor + Grok Bot) usage instead of 2x Cursor with Grok Bot consuming from the same quota.
Grok Bot now using Opus 5.5 seems to burn through the weekly quota faster than before.
Some folks want Other Models, and that’s fine. But even for folks like me who are okay with only Grok models, the quota doesn’t seem to have improved for the $30 X plan compared to $20 Cursor plan.
If the X team can increase the weekly quota to account for 2x (Cursor + Grok Bot) usage or at least a bare minimum of 1.5x (Cursor + Grok Bot) usage (assuming 1.5x from $20 to $30, and similar for other plans), that would make these plans comparable or better.
Honestly this new X Ai plan sucks, I need the old cursor legacy plans.
Regarding emailing [email protected]. I sent the email on 10/15/26 and no response yet. Does anyone know when I can expect the response? (I’m not talking about the AI response that comes back in a minute. The last I’ve heard is “I’ve passed this to a teammate who can help. They will take a look and follow up here soon.”)
Hi there, i am trying to upgrade my seat and its not letting me - has anyone else experienced this problem before?
Ticket T-H43741. I’m still on legacy Pro Plus and considering Super, but after three rounds with support I still can’t tell you how the 2.25x figure is calculated. Not in tokens, not in dollars, not in requests. They also wouldn’t confirm whether my separate Grok Bot pool is included in it. If the support team itself can’t explain the math behind the number on my checkout screen, how am I supposed to make an informed decision? That’s not transparency, that’s a black box with a price tag.
A few things I’d like the team to address directly:
One — publish the pool sizes in absolute units. Even a rough range in dollars or tokens per tier would let people compare old and new plans on equal footing. Right now “15x Plus” means nothing concrete to someone who has been tracking their actual consumption.
Two — clarify the weekly reset against monthly legacy allowances. Colin confirmed the 2x is spread across weeks, which effectively means each week is roughly half of what a monthly Pro plan gave me. That’s a fundamental shift in how bursty development work fits, and it wasn’t obvious from the announcement.
Three — state plainly what happens to Grok Bot when it shares a pool with Cursor and the Grok apps at different burn rates. One user on this thread reported 1.5 billion tokens on Cursor Models plus 2 billion on Grok Bot in a single month, and ran out of Cursor Models while still having half a Grok Bot week left. That asymmetry is exactly why separate pools existed.
Four — bring back any included third-party allowance, even a modest one. Without Claude or GPT in the mix, Cursor’s own models aren’t sufficient for the long-horizon planning work a lot of us do. Several people on page ten said the same: no Other Models pool means no point using Cursor for their projects.
I’m not switching. Not out of nostalgia, but because the only tier that resembles Pro Plus costs seventy percent more, gives me a shared weekly pool I can’t size, and drops capabilities I was already paying for. If the goal is migration, this pricing does the opposite — it makes staying put the rational choice, and it makes the “you can always switch later” framing feel one-directional.
Is it possible to use China’s Alipay?
In China, users must use a VPN to bypass internet restrictions in order to subscribe. Are you aware of this?
That was a fun ride. Owner of 4 Pro Ultra subscription here. To be said, this is a pure joke. Great way to loose honest reliable customers. Sending my regards to Elon.
I just switched from cursor Pro to X subscription but it’s not migrated my project and data from cursor and I’ve already paid! What going on? How do I fix this?
Thanks
Has anyone tried replacing Cursor/Grok Bot with Hermes and one of the open models?
The usage problem on Grok Bot made me research it and I’m considering installing a model on my local machine. I asked Gemini and it recommends:
- DeepSeek 4
- Meta Llama/ollama
- AliBaba Qwen/Qwen-Coder
- Zhipu/Moonshot GLM/Kimi
- Google Gemma 4
Any I’ve missed to consider?
I haven’t tried any of them yet. I’ve also never used Claude Code or ChatGPT Codex. I’ve only ever used Cursor and Grok Bot.
Thoughts on how Hermes compares to Grok Bot for the chat, and on how the open models compare to Grok for coding?
Gemini also recommended the following Grok Bot alternatives:
- OpenManus/OpenMausBot
- Rakazo
- AutoGPT & BabyAGI
- Open Interpreter
- LibreChat
Thoughts on any of those, or other alternatives?
This is so bad! I don’t use any grok or X, I had 350$ of credits because people signed up with my referal and now I just needed a basic plan to use various models.
Bought Plus plan and its just has grok models with only medium effort version. Nothing more. This was nowhere written… Maybe warn us before the change or specify which sub has which model…