Developers who use Google’s AI code tool Antigravity are critical of changes to the pricing structure of Google’s AI services. According to many users, the available usage limits have changed, making it harder to use the platform without incurring additional costs.
This is reported by The Register. Google recently announced that it is adjusting its AI subscriptions to give users more flexibility in how they use AI tools. In the new model, so-called AI credits play a central role. These credits can be used for Antigravity. Subscriptions include a number of credits, while additional credits can be purchased separately. For $25, users receive 2,500 credits. Exactly how much one credit represents in terms of usage with Antigravity is not clearly specified.
The announcement led to many responses on the Google AI for Developers forum. Users with an AI Pro subscription in particular report that their experience has changed significantly. According to the subscription description, users receive a generous usage quota that is regularly renewed until the weekly limit is reached. In practice, developers report that this renewal occurs less frequently than before.
Several users indicate that their available quota is no longer replenished every few hours, but only after a week. As a result, projects are temporarily halted unless they purchase additional credits or switch to a more expensive subscription.
Antigravity supports several large language models. These include variants of Google Gemini, as well as models from Anthropic and OpenAI. The most efficient and cheapest model in the range is Gemini Flash, which, according to Google, is intended for lighter applications.
Subscriptions for different types of developers
Google positions the AI Pro subscription primarily for hobbyists, students, and developers who primarily work in their development environments. For more intensive professional use, the company refers to the more expensive AI Ultra subscription, which costs almost $250 per month and is intended for higher-capacity access to more complex models.
Some developers closely track their usage and claim that the limits have been significantly reduced. Whereas hundreds of millions of input tokens per week were previously possible, they now report that the limit is already being reached at a fraction of that.
Antigravity was introduced as a preview in November 2025. At launch, Google did not provide clear information about the final pricing structure. Instead, the company referred to its AI subscriptions and described the available quotas in general terms without specific figures.
Complaints about unclear limits and unexpected restrictions have been circulating within the developer community for some time. Some users report that their workflows are interrupted by sudden quota exhaustion. They are asking Google to be more transparent about how usage limits are calculated and why available capacity sometimes seems to disappear faster than expected.
The cost structure of AI services remains complex. Running large language models requires substantial computing power, and it is difficult to predict how many resources a prompt will consume. This makes it difficult for providers to maintain stable pricing models.
Google has not yet explained in detail what the recent changes mean for the use of Antigravity. It also remains unclear how much processing capacity an AI credit actually represents within the platform.