I’ve been a Google AI Ultra for Workspace ($250/month) customer for the past several months. Recently, Google removed Google AI Ultra for Workspace users and instructed customers to move to Google AI Ultra on personal accounts if they wanted to continue using Antigravity, since it’s being removed from Workspace.
I made the switch today, and the difference in rate limits is insane.
On Google AI Ultra for Workspace, I could use Antigravity with Claude Opus 4.6 continuously for nearly two hours without any issues. On the personal Google AI Ultra plan, I’m hitting rate limits after roughly 20-30 minutes of the exact same usage, sometimes even less.
This is supposed to be the Ultra plan, yet I’m being rate-limited this aggressively and then told to wait several hours before I can continue using the service.
What’s frustrating is that this was rarely an issue during the months I was paying for the Workspace version. What’s also disappointing is when I previously contacted Workspace support about an issue I was having, I was told to use Google’s models instead of Claude.
At this point, it’s difficult to justify paying $200/month when I could subscribe directly to the model providers themselves. The primary reason I subscribe to Google AI Ultra is Antigravity. I have little interest in most of the other features being marketed with the plan.
If Google offered an Antigravity-focused subscription with substantially higher limits, I’d gladly pay for it. But the current experience feels like a significant downgrade from the service I was already paying for.
I agree and am looking for a workflow that will match the Antigravity 2.0 experience. The moment this happens, I am moving on. Google, for once was a head of the curve, but their inability tell us what is happening and provide us with transparency has soured me on sticking around. While I am retired and doing this for fun .. I find myself struggling with the process more than the coding when Antigravity is involved.
I need somethig that just works and (regardless of how expensive) tells me how much I spent and on what exactly.
This. I’ve been looking and looking trying to figure out what went wrong and where my usage is going and I still can’t figure it out and that’s a headache. If I understood why my usage is the way it is now I’d probably be more understanding but the issue is now I’m using even LESS than I was on Google AI Ultra for Workspace and now on my personal Google AI Ultra subscription my usage has completely depleted in 1/4th the time.
Antigravity 2.0 is great, even the IDE (which I’m primarily using), and Antigravity 2.0 for the agents but the usage is honestly disappointing.
My workflow was essentially getting Flash 3.5/3.1 Pro to handle the groundwork with Opus fine-tuning it but even now that’s impossible because of the insane difference in usage with personal Google AI Ultra…
I was accepting of the occasional spikes with the models at max capacity and waiting but seriously now it’s completely unusable for my usecase.
We’re all in the same boat. I liked the idea of managing my AI subscription through my workplace account. Both for auditing and management reasons.
Google has made this impossible - AG doesn’t even recognize the new workplace AI extra subscription. It’s all broken.
I had to shift my workflow to Curser+Claude in order to avoided the work disruptions. It’s actually quite nice. Claude desktop is handy for some non-dev work and Curser has the agentic/IDE interface.
Hilarious story. I asked Gemini how software dev companies on workplace accounts manage their AI with Google - Gemini says they don’t, everyone uses Codex and Claude.
