Free Gemini users now on ‘Auto’ models as app adds thinking levels

Under the newly deployed system, users on the free plan—officially categorized as those without an active paid artificial intelligence subscription—will no longer have the ability to manually select specific models from a drop-down menu. Instead, the model picker is now restricted entirely to an "Auto" option powered by the Gemini 3 model family. According to Google’s updated support documentation, the backbone of this automated selection process relies heavily on the Gemini (3.5) Flash-Lite model, designed to maximize speed and efficiency for everyday queries.

Google explains that the new auto model selection architecture dynamically routes each individual user prompt to the specific Gemini model deemed most appropriate for the task at hand. While the vast majority of standard queries and conversational prompts will be handled seamlessly by the lightweight Flash-Lite model, tasks requiring more complex computation, deeper analytical reasoning, or advanced problem-solving can be automatically redirected to more robust variants such as Flash or Pro. However, users who prefer to opt out of the automated routing mechanism retain a measure of control; if smart model selection is disabled within the application preferences, chats for free-tier users will default exclusively to Flash-Lite for all generated responses.

Alongside the changes to the free-tier experience, Google is introducing a new suite of "effort levels" across the platform. These newly introduced settings officially replace the previous "Extended" option, aligning consumer-facing applications more closely with the advanced thinking levels previously found in Google AI Studio and Antigravity environments. This adjustment gives users greater direct influence over how much computational depth and processing time the AI applies before delivering a final response.

The rollout is happening rapidly across the ecosystem. Observations from late in the day in the Pacific Time zone indicate that the model selection changes for free accounts have already hit the vast majority of test and user accounts, demonstrating a swift deployment schedule by Google’s engineering teams. Similarly, the newly introduced effort levels are currently in the process of rolling out to all eligible subscribers globally, ensuring a consistent feature set across various tiers of service.

The restructuring of the Gemini application ecosystem is not limited to free accounts. Additional adjustments are slated for paid tiers as well, with Google AI Plus users soon facing refined model parameters that restrict their access primarily to the Flash-Lite and (3.8) Flash iterations. These ongoing adjustments highlight Google’s broader strategy to fine-tune its product segmentation, balancing computational load and operational costs while delivering optimized performance tailored to specific user tiers.