Agreed, the new limits are not making sense.
Thatās crazy, tell me how itās possible to provide an AI coding Tool with this system of long wait ?
i think they doesnt care about people. Tips : just provide 10 dollars to OpenRouter, and you will gain 1000 requests per day for free, without spending this dollars.
I enjoyed antigravity, but, they killed their tool.
+1
I genuinely donāt understand how such a poor idea could possibly be conceived; itās like youāre paying for a demo of a demo. I was already falling out of love with Antigravity when everyoneās usage vanished into the ether earlier this year. It was at least useable after that though. Currently itās an absolute joke; why am I even wasting storage to start something, fail to finish it, and then wait X period of time? Like I said, demo of a demo.
Regardless; Codex, Cursor and Devin (formerly WindSurf - but havenāt tried post conversion) blow this out of the water. The latest Codex trial with double rates on the high tier package has been magnificent.
Also, I have to say, the absolutely unnecessary complexity and convolution of Google Cloud and the 50 versions of Gemini product is also an absolute nightmare. God awful. I feel like every day I find out thereās another Gemini related product. And god forbid you try and find useful, comprehensive, or clear information on the various heads/payment plans/APIs/features of the Gemini//Google hydra.
Bring it back. Donāt make it worse.
+1 it seems like we need the 3 days quota of every model separate from each other so we decide which one to use
Iām coming from Gemini CLI and itās insanely amazing as a user. The AI āāGoogle Pro plan has a quota reset every 24 hours, but it lasts a long time. You get the 3.0 Flash/Pro model and the 3.1 model, and it has automatic routing that routes your requests to the model based on your task.
My Gemini CLI plan expires on June 18th, and Iāll have to use Antigravity. Iāve been using it for a few days, and itās the biggest piece of junk, with all due respect, but it has many flaws.
-
The models share quotas, while in Gemini CLI they donāt. That is, you finish the Flash 3.5 model and you canāt use anything else; you run out of all the models.
-
You cant use the old models GEMINI 3.0FLASH/PRO in Antigravity while in Gemini CLI yes.
-
The daily quota in Antigravity with the Pro plan seems to last much less than in Gemini CLI.
-
Yesterday, with the update, they introduced a weekly quota. I find it unbelievable coming from Gemini CLI, even though itās the same companyā¦
I really donāt know where else to complain or share this feedback because it seems like Google isnāt listening. Itās tedious using the models in Antigravity when itās 100 times worse than in Gemini CLI. I honestly donāt know how many requests Iāve received in 5 hours of resets, and I see that my project is running out incredibly fast. Supposedly, the 3.5 FLASH model should consume much less than the 3.1PRO/FLASH model, but I donāt get that impression.
+1 My experience is still worse after this update
The weekly limit is absurd and needs to go. The trickle needs to be much more even. Iām fine with dipping into overages every now and then; with a predictable refresh of the base quota of a paid plan, it is straightforward to pace oneās work to both get stuff done every day and fit a budget.
(PS: I do not see the downsides of the 3.5 Flash models; theyāre different yes, the large context window is great. I tailored a GEMINI.md file to make the agent be conservative and parsimonious; that works for me, Iām getting more output of higher quality at the acceptable cost of a bit more of my own review effort.)