Unexplained Quota Consumption and Lack of Transparency in Antigravity Pro Plan

What is happening with the Antigravity service is, at the very least, unacceptable for a paid product that positions itself as a professional solution. The basic expectation of any user investing in a Pro plan is predictability, transparency, and consistency in delivery. None of these are being properly met.

Recently, beyond the noticeable reduction in quota without any official communication or technical justification, an even more serious issue has emerged: resource consumption without any user action. In the scenario described, the quota was renewed after the expected waiting period, returning to 100 percent, and without any commands being executed or any interaction taking place, it immediately dropped to 40 percent, along with alerts about future limitations. This is not just an inconvenience, it is a direct breach of trust.

Quota management is a fundamental pillar in consumption-based services. When this control fails, or worse, appears to be applied in an arbitrary and opaque manner, the user completely loses the ability to plan how to use the tool. It becomes impossible to operate in a professional environment where resources are drained without traceability, without clear logs, and without any accessible auditing mechanism.

Additionally, the silent reduction of quota over time reinforces the perception that the service is deteriorating. Changes of this nature should always be accompanied by clear communication, explaining the rationale behind them, the impact on users, and any available alternatives. The lack of transparency signals disregard for paying customers and undermines the platform’s credibility.

Another critical issue is the lack of operational predictability. A service that shows passive quota consumption raises serious concerns. It suggests the possibility of background processes running without consent, flaws in session control, or even errors in the usage accounting mechanism. Without clear and objective answers, users are left dealing with a system they cannot trust or control.

For a Pro plan, this situation is even more problematic. Payment implies not only expanded access but also stability, reliable support, and proper governance over resource usage. When the service becomes more frustrating than productive, the cost can no longer be justified.

Unless there is a prompt resolution and, more importantly, transparent communication about what is happening, including detailed usage logs, technical explanations, and guarantees that the issue will not persist, dissatisfaction will continue to grow. Paid services cannot operate under uncertainty, and in its current state, that is exactly what is being delivered.

I totally agree. I am very disappointed with what is going on. It is taking a matter of 30min to burn through quota now on my AI Ultra Plan. Completely ridiculous and not remotely good value for money. I’ll prob just jump ship and head on over to Claude ASAP if this isn’t resolved. It this another potentially dead project by Google? Such a shame, I quite like Antigravity. Antigravity Devs, please fix this ASAP if it is a bug. Otherwise as others have pointed out, be transparent so all those handing over mega dollars for AI Ultra plans (and other plans) can decide if they want to continue paying Google to get next to nothing in return.

I am primarily the non Google model user. I have Google AI ultra. Antigravity used to work really well when I started a month ago, first 15 days were honeymoon. Then in 3 hours that quota got exhausted but it refreshed in 1 hours~. Fine, I used it as a lunch break or to think more creatively, working on my prompts. I was able to save my included 25000 credits and used them in last 2-3 days before first billing month ended.

In Second billing month, all my included credits consumed in first 4 days. Model Quota for Claude Opus 4.6 expires in less than an hour. Even if I am working on a singular task, keeping my chat history small. Not to mention annoying retry rate limit alerts.

In Ultra plan, I notice generous limit for Gemini models. However sometimes they require multiple correction loops and force me to function like a micro manager where I tell it what function to work on, how to write code in that function.

So, I might as well switch to the company who built the coding model, I love. I did not need the bells and whistles of Google AI ultra anyways.