The state of the "Antigravity Ultra" plan is a joke

Let’s talk about the recent silent changes to the Anti-Gravity Ultra plan, because this is getting absolutely ridiculous. The operational mindset behind this is completely flawed.

1. Stealth-nerfing Claude quotas: It’s glaringly obvious that you guys are secretly slashing the usage quotas for the Claude series. No public announcements, no transparency, just silent downgrades. Doing this to paying users isn’t just terrible business ethics; it’s practically a scam. We paid for a promised service, and silently pulling the rug out from under us is unacceptable for any serious company. It crosses the line into legally questionable territory.

2. Force-feeding us an inferior model: While Claude gets nerfed into the ground, the quota for the Gemini models remains suspiciously generous. We all see what you’re doing: trying to force us to use your preferred model. Here’s the fatal flaw in your master plan—the model you’re pushing is basically brain-dead. Its performance is a joke compared to its industry competitors. Trying to shove a heavily inferior product down our throats is a desperate, backward move (seriously, what are you thinking?) that will only drive your user base away.

3. Wake up and fix it: The management team needs to take this seriously before you ruin the platform completely. You have two options:

  • Option A: Restore the originally advertised quotas, honor your promises, and actually respect your paying users.

  • Option B: Put in the actual R&D work to make your models competitive, so they aren’t embarrassingly far behind the industry standard.

Stop relying on these unethical, shady tactics to boost usage. We aren’t stupid, and we notice.

34 Likes

This. Every single point.

Multiple users have already shared their numbers. The pattern is undeniable. What you’ve laid out isn’t speculation — it’s exactly what we’re all experiencing, just said plainly.

Option A isn’t a demand, it’s the minimum.

4 Likes

It’s pretty obvious they are trying to funnel us into using the Gemini series, but the performance gap between their models and Claude is simply too wide. Following this trend, if their models can’t actually compete with the Claude lineup, this product is basically doomed. In an era flooded with AI tools everywhere, who is going to use a sub-par, incapable model for professional development work? Personally, I think Gemini is fine for casual chatting, but for actual coding, it’s a hard pass.

What’s truly appalling about this platform, however, is the complete lack of transparency regarding usage quotas. If they were just upfront about the limits, users might even be willing to pay a bit more for what they need. I genuinely have no idea what their product team and management are thinking.

9 Likes

If thats the case then transparency on the actual quota is the way forward. Publish how much quota do Max and Ultra users have and keep decrementing the use so folks can see it.

Current system seems bogus now.

1 Like

I agree with this completely.

What is making users angry is not just the quota reduction itself, but the apparent lack of transparency around it. When paying users subscribe to a top-tier plan, they do so based on the practical access conditions being offered at that time. Quietly reducing that usable access, while keeping the same premium price, is exactly why so many of us feel misled.

That is the core problem: the service may still carry the same plan name, but the real-world value and usability appear to have been reduced in practice.

I am personally very frustrated by this as well. What we are seeing looks less like a normal product adjustment and more like a silent downgrade imposed on paying users without proper notice. That is not acceptable.

And when users then raise legitimate concerns, the responses have often been generic, evasive, or framed in marketing language instead of directly addressing whether access conditions were actually changed.

Paying users are not confused. We can tell the difference between the plan label staying the same and the actual usable service getting worse.

At a minimum, there should be:

  • full transparency about what changed,

  • restoration of the original practical access conditions, or

  • a price adjustment that reflects the reduced value.

Users are right to be upset.

4 Likes

Everyone is freaking out on Twitter because Claude adjusted quotas, but they know that because Anthropic put out a statement explaining the quota adjustment. I’m like “you are all so lucky, Google just changes quotas and says absolutely nothing”.

3 Likes

It is a sad state for Antigravity. I upgraded to the Ultra plan because I was reaching limits with Pro.

And I feel that I get even less usage than what I had in Pro now, and this flipped in the last 1-2 days.

Also gemini-cli became unusable, getting “no available compute” and “rate limited” all the time.

No idea what Google is doing with developers.

1 Like

Yes see it right now. In Ultra Gemini CLI is unusable, stating: “This is taking a bit longer…”

There are no SLAs or SLOs for Antigravity and the Gemini CLI. The tools are provided to you “as is” for developer use. If you have followed the Gemini CLI announcements recently, you know they nerfed all OAuth users, and you should use the API pay-as-you-go method for higher rate limits and a better user experience. AI Studio now supports budget caps in case you burn through $300 during your usage.

I’ve officially jumped ship to Cursor and Claude Code. Honestly, I started out as a Gemini user, but after trying Claude, I realized its coding abilities are just leagues ahead. Now that they’ve gutted the Ultra quota, I basically burn through it in like 2 to 4 prompts. I tried falling back to Gemini 3.1 Pro and was massively disappointed—the speed, the coding logic, it’s not even close. I think it’s time to pull the plug on my Ultra plan. You win, Google. Let’s pack it up folks, nothing left to see here.

5 Likes

I will go for Codex as they have released a Windows app quite recently.

I am still an ULTRA user, but my Claude quote is empty after 5 prompts… So, yeah…

3 Likes

Hello everyone!

Thank you for bringing these concerns about quota limit in AI Ultra plan to our attention. Please be assured that I have shared your feedback with our internal team for further review.
We appreciate your continued patience as we work to enhance the Antigravity experience.

2 Likes

Broken record (boys who cried wolf) - there is a cultral issue within your company around pre-notification of any customer impact changes.

Is Gemini-2.5-pro disabled for free-tier? - #19 by J_J1

4 Likes

The recent changes just killed everything why Antigravity was interesting. I cancelled my Ultra subscription immediately as the actual quota changes are rendering the solution useless. Gemini 3.1 as a model is not sticking to the Agent Harness and the output is not on par in comparison to Claude Opus. Being able to work only for 30-40 Minutes with one model and around 2 hrs with the other and then wait for hours is just way overpriced. Google was on a streak and now again lost it.

The weeks before, I never hit a quota limit with a great amount of work being done; this was what I expected as a customer.

3 Likes

Yes, JUST A … . . . . .

It’s really unfortunate, but I’m about to cancel my Ultra plan as well. I can’t justify spending $250 a month when I get somewhere between 10 and 20 minutes of usable time, and I’ve done a lot to reduce my token usage. I’m extremely frustrated with Google. This is not acceptable, and they need to come out and at least address the issue. All my projects are about to move to Claude or Codex.

3 Likes

I just started experiencing this today. It’s like they realized you favored Opus over Gemini for coding and decided for you that you should be using Gemini now. While I do like Gemini for a lot of tasks. It’s current state with Antigravity IDE is pretty … to say the least.

It’s not a joke. It’s completely intentional. By design.

2 Likes