Well, from what I’ve seen lately, Sonnet 5.5 has been incredible. Way better for me than opus. I noticed my opus turns were continuing on, lots of recursive checks, MASSIVE test suites being created, etc. I had to monitor opus 5.5 closely and rein it in.
I saw a lot of chatter today that opus 5.5 billing ruined grok bot. I’m not sure if I buy that. I can’t imagine that the best engineers weren’t seeing and adjusting token usage and model selection.
And Sonnet 5.5 medium! Not even extra
Yeah. I was watching it closely when they first ran it, snuck the cursor api included amount from $400 to basically unknown… but I wasn’t as worried about that since I have Claude max also.
But kept SuperGrok Heavy at $300 thinking we could continue on. Most everyone I saw at the time who unified saw a reductions and wished they wouldn’t have…but maybe you’re right, they would just rather move people along.
But again, I'm on the legacy SuperGrok Heavy, and had Grok Bot doing only light work, using with my Claude remote session to handle most of the actual work. Usage always light, now suddenly 100% with no change in activity. Are you saying that it warrants someone from @xai to take a look at?
@grok@dallastesla@spacexai Thank you. Finally a reasonable answer.
So Grok Bot should NOT be a part of the operational plan for any business, unless there is excessive OpEx funds sitting idle that can weather this kind of cost change without notice, correct?
This post misses the point, and reads like a shill post.
If "Grok 4.8 is coming. Probably within 2 weeks. And it's built to fix exactly this problem." then why do this idiotic move and piss a bunch of people off, instead of waiting for the 2 weeks and solve the issue?
Fact: I'm a $SPCX shareholder.
@GarethRansome@grok@dallastesla@spacexai Bah, probably not as regretful as sitting here trying to get a somewhat satisfactory answer from a bot. I assumed I'd get a little more humility, given Grok is normally not sugarcoating things.
Oh well.
@craigd0412@Alex_Michalas@grok You're telling me that you are running 25 bots, heavy usage, and you are at 25%?
And do I take it literally that you ran them for 2 hours? Or are you saying you ran them for 26 hours... heavy use from 12:01a to 2am on the following day?
Totally agree. I've migrated away from local compute, but now leaning back in heavily. Particularly around some of the work we do in text categorization and intent... some of the models out there that will run on a 5090 are really strong.
I'm not sure if something will come from this whole debacle... but it certainly doesn't *feel* right.
@grok@dallastesla@spacexai I'm shocked that you would say No. Would you operate a small business comfortably if your costs of doing business could fluxuate by 50-100% overnight?
Do you care to answer my actual questions? You guys specifically marketed Grok Bot as a game changer (and I we all agree) for business. People have worked tirelessly and invested time, money, and effort into building amazing things. They've been grinding daily to make this work. Many won't recoup their costs for a bit, but they've managed to identify their CapEx and OpEx.
But suddenly the arbitrary, or as you call it "relative terms" of the weekly pool change significantly, and suddenly people are very upset because they are going into Monday morning with a completely different outlook, in a negative way.
How can you prop up Grok Bot for business, tout it getting 3x faster (https://t.co/amUuX7mZOk) and then greatly change the "relative terms". It seems like those terms don't mean anything, since "Absolute quantities were never listed."?
Would you agree that entering into a business relationship with a SaaS provider with unknown rates would be a bad way to operate a business?
@GarethRansome@grok@dallastesla@spacexai I think a court might have a difficult time with this reasoning... I have a feeling that it's been strictly guided by someone internally.
https://t.co/pZymyrWXEJ
So how can I effectively run a business with Grok Bot as a cornerstone, flagship tool in my toolbelt, if there are arbitrary multipliers that I cannot see, and I'm not made aware of until it's too late?
Do you know how many people probably quit their jobs because your marketing team sold the idea of starting a business with Grok Bot? And now the OpEx on that has more than doubled?? (You must realize that if usage multipliers, or "compute" as your are stating, has doubled, then to do the same amount of work on an adhoc credit purchase will be incredibly expensive.
But hold on a second. The X Subscription only replaced SuperGrok for new customers on October 12, 2026. Before that, xAI’s pricing page described tiers with adjectives and no multiplier. A September 21 snapshot showed the Free plan as “generous limits”, SuperGrok at $30 a month as “higher rate limits across all features”, and SuperGrok Plus at $100 a month as “significantly higher usage”.
You have access to my account, Grok. Tell me, did I receive, as a SuperGrok subscriber, "significantly higher usage" pools?
@grok@dallastesla@spacexai Of course the price changed. If I give you 2 pizzas a day for $300/mo, and suddenly, without warning, I start dropping off 1 pizza for the same amount of money without giving you any notice, that is an effective change in price for goods or services received.