Google is gradually adjusting its AI model usage policies. Since May, the Google AI Studio platform for developers has stopped providing free users with access to call the Gemini2.5Pro series models. The official response stated that this is part of the transition to a new system based on API keys, and in the future, developers will need to provide their own API keys to access model services.
During the current transition phase, Google still provides free API keys based on Google Cloud for developers, and emphasized that the free package of Google AI Studio "will not disappear in the short term." However, many developers have found that when trying to use Gemini2.5Pro, they encounter an HTTP429 error message, clearly indicating that this series of models does not support free tier calls.
Currently, free users can still access the Gemini2.0 series models, but in terms of capabilities and performance, this series lags behind the latest Gemini2.5. This policy adjustment has drawn widespread attention and discussion within the developer community, especially impacting users who rely on Gemini2.5Pro for high-performance testing and prototyping.
The employee responsible for Google's AI API platform responded in the community as follows:
"We are still providing API free packages for millions of developers, and the 2.5Pro model can still be used for free in the Gemini application. In the face of high demand, we have provided double the free computing resources and are exploring possible paths to continue opening up free versions in the future, such as setting lifetime call limits or launching incentive programs."
However, he also admitted that whether free access to models is allowed is subject to various limitations, particularly the cost issues of the models themselves.
Analysts believe that Google's move may aim to promote the commercialization of Gemini, incorporating more high-performance model usage into the paid system, while maintaining the user experience on the Gemini application side. For developers accustomed to "free-riding" on AI Studio, they will have to make choices between performance and cost in the future.