When discussing artificial intelligence, it's essential to differentiate between various terms. While we often refer to "intelligence," the concept of "thinking" has emerged, particularly in models that process requests in multiple stages, leading to a more deliberate output. Gemini, Google's AI application, has now introduced new "thinking levels" for users engaging with non-thinking models.
As reported by 9to5Google, the Gemini app for both iOS and Android now features a "Thinking level" option in its model selector. This is distinct from the established "Thinking" model within "Gemini 3." Instead, the "Thinking level" becomes available when utilizing the Fast option, which employs Gemini 3 Flash, or the Gemini 3.1 Pro model, provided you are not currently using the Thinking model. Users will encounter two choices: "Standard" and "Extended." According to Google, the Standard setting is suitable for most inquiries, while the Extended option allows for deeper processing of more complex subjects.
This enhancement aims to provide users with greater flexibility in processing their requests. For instance, if you opt for the Fast model, you might find that the answers generated are not as comprehensive as desired due to the emphasis on speed. By selecting the Extended Thinking level, users can invest a bit more time but potentially receive a more satisfactory response, all without resorting to the full Thinking model, which may require additional time.
Notably, this feature does not appear to be limited to Google AI Plus subscribers, as evidenced by the screenshots shared by 9to5Google. Free users should also expect to see the "Thinking level" option becoming available soon.
These updates come just ahead of the Google I/O event, where the company is anticipated to unveil further advancements in Gemini and other AI technologies. Recently, Google hosted The Android Show: I/O Edition, highlighting several Android innovations before the main event, indicating a strong focus on AI developments during the upcoming I/O.
