Hi Antigravity team,
I subscribed to Google AI Ultra pretty much the day after Google I/O and initially loved it. It was the first platform where I could genuinely get a full 8–10 hour day of deep development work done without constantly fighting limits.
I took a break for the prior two weeks. Coming back in the last 24 hours, after just two 2-hour work sessions on the first day of a new weekly cycle, I’m already at:
- Gemini Models: 77% of weekly limit used
- Claude & GPT Models: 60% of weekly limit used
This is despite the 5-hour rolling limits still having plenty of headroom. I’ve gone from being able to do full, uninterrupted days of work to barely managing 4 productive hours before the quotas start seriously constraining me.
For comparison, on my separate Claude Max subscription I regularly run 3–4 extended sessions on Opus 4.8 for up to 18 hours a day — and I still only just approach my weekly limits. It actually takes a very heavy workload to max it out.
The token economics on Antigravity appear to have shifted significantly for the worse since I first subscribed. The current weekly limits feel overly restrictive for an “Ultra” tier, especially when compared to dedicated high-tier plans on other platforms.
I have a few questions:
- How is quota consumption calculated across the different model families (Gemini vs Claude/GPT)?
- Are these limits final, or is there an adjustment planned for Ultra subscribers?
- What usage pattern is the Ultra plan actually designed for?
I’m a heavy power user focused on AI agent orchestration, research, and development workloads. I was genuinely excited about the unified access Antigravity provides across frontier models. Right now though, the quotas are making sustained deep work very difficult.
Happy to share more details or additional screenshots if helpful.
