Google to Limit Model Selection for Free Gemini Users

    Starting October 9, Google will limit model selection for free users in the Gemini app. Users without a Google AI subscription will only be able to use the Flash-Lite 3.5 model. From that date, they will lose access to the Flash 3.6 and Pro 3.1 models. In May, the company changed usage limits due to the computational load.    AI Plus subscribers who pay $4.99 per month (approximately 8 manats) will also have fewer options. Flash-Lite and Flash will remain in the package, but the Pro model will no longer be available. Google will notify subscribers by email of the exact timing of the changes.    No reductions in features are planned for the AI Pro and AI Ultra packages. On the contrary, the Deep Think option will be added to the AI Pro package, which costs $19.99 per month (approximately 34 manats). This option, which processes several lines of thought simultaneously, is currently available only in the $99.99-per-month (approximately 170 manats) and $199.99-per-month (approximately 340 manats) packages.    In October, thinking levels called low, medium and high are expected to be introduced for each model in the Gemini app. At the high level, tasks will be processed in greater detail and responses will be more comprehensive. However, the user’s query limit will be used up more quickly in this case. Currently, only the Extended thinking option can be enabled in the app for complex problems. The same three-level selection is already available in Google AI Studio and Antigravity.    It has not yet been announced which tier the new Gemini 4 Argon model will fall into under this structure. Google has said it will initially make the model available to AI Ultra subscribers. However, it is unclear whether Argon will be classified as Pro or included in a new, higher-tier package. It has also not been clarified when the other packages will gain access to the model.