Antigravity Pro has lost most of its value

I’m struggling to justify $20/month for Antigravity Pro right now.

The biggest issue is that Gemini 3.1 Pro and Gemini 3.7 Flash are gone. They did different jobs in my workflow, and having several models with different strengths was a big part of why Antigravity was worth paying for. Now it feels like Gemini 3.8 Flash is the only model I can count on.

I know the usual answer: “If you need something stronger, just use Claude.” I’d love to, but the usage limits make Claude hard to rely on for serious, sustained development. What happens when I hit a complex task, burn through my Claude quota, and have no capable Gemini Pro model to fall back on? At that point I’m looking elsewhere for my hardest work anyway, so why not move to Claude Code or Codex entirely?

Getting to a genuinely powerful model like Argon means upgrading to Google AI Ultra, which costs about five times as much. At that price, I’d rather pay for Claude or ChatGPT/Codex, which come with their own coding workflows and a solid range of models. I don’t see Ultra as better value.

I’m not asking for unlimited access to the most expensive models. I’m asking for a Pro tier that gives me a usable lineup for the different stages of development:

  • A capable Pro-level model for complex reasoning and implementation
  • A fast model for routine tasks
  • Reasonable usage limits on the alternative models

Right now I’m paying for a service that leaves me depending on one model that doesn’t cover everything I need.

So please do one of the following: bring back Gemini 3.1 Pro, offer a capable replacement for Pro subscribers, or give us a reasonable path to Argon without a fivefold price jump.

Otherwise, what exactly are we paying for?


GIVE GEMINI 3.7 FLASH OR GEMINI 4 ARGON IN PRO PLAN IN ANTIGRAVITY IDE @Goo.gle @Gemini_AIS_Support

There is NO Argon in the Pro plan. One must move to Ultimate, a 500% increase in price. Even Gemini 3.8 Flash High running from inside Antigravity said it’s not worth the $ and recommended switching to Cursor and using Claude Opus/Sonnet directly from there.

downgraded to ai plus .. having ai pro without a pro model for coding is joke. i will try codex or just stay with my company claude access this one is finding errors even in the simplest code snippets generated by google .

its not antigravity that lost value. your issue is because you didn’t follow one of the architecture rules stated in the SOLID concepts. you hand over to a model the ability to break your system, the model depend on another company that you cannot control. this is a issue from lack of experience! solid software shoud not rely on a blackbox. a change in the temperature output on the server side to save money, it will make your model not usable in production and despite you will burn many token.