Free Gemini is not going away, but its choice of models is about to get much narrower. From October 9, personal-account users on the free tier will be limited to Flash-Lite, while access to Flash and Pro models will depend on a paid subscription.
How model access changes
Google is assigning access by subscription tier, not by a particular generation of Gemini. The current examples include Gemini 3.8 Flash and Gemini 3.1 Pro, but future model updates will not, by themselves, change the tier structure.
| Subscription | Model access |
|---|---|
| Free | Flash-Lite |
| AI Plus | Flash-Lite and Flash |
| AI Pro and AI Ultra | Flash-Lite, Flash, and Pro |
The practical result is straightforward: trying a more capable model will increasingly mean choosing a paid plan. The free tier remains, with a more limited model selection.
AI Pro adds Deep Think
Google is also adding Deep Think to AI Pro. The company describes it as a mode for “maximum parallel reasoning.” It was previously one of the features associated with the more expensive AI Ultra plan.
In October, users will also be able to select a low, medium, or high compute level for a model. Higher settings let the model use more computing resources to prepare an answer, but they also use the user’s available quota more quickly. More effort from the model, in other words, comes with a larger draw on the meter.
The limits of the change
These adjustments apply to Gemini’s consumer apps for personal accounts. They do not automatically extend to the Gemini API, Google AI Studio, or Google’s other products.
Google is separately changing access to third-party models in its Antigravity development environment. Claude Sonnet 5.5 and Claude Opus 5.5 will be available to AI Pro and AI Ultra subscribers. Free users and AI Plus subscribers will not have access to them.
The new setup makes the dividing line less about which Gemini generation is current and more about which plan a user pays for. For free users, the answer from October 9 is Flash-Lite.