Google has completed the rollout of its “Thinking Levels” feature for the Gemini app, making Standard and Extended reasoning modes available to all users across web, Android, and iOS platforms. This development represents a meaningful step in Google’s ongoing effort to give users greater control over how they interact with its artificial intelligence systems, empowering individuals to tailor the AI’s reasoning approach to the specific demands of their queries. Josh Woodward, VP of Google Labs and the Gemini app, confirmed the wider availability on X on Tuesday, June 2, writing: “Papercut fixed: Thinking Levels are now available on Gemini across Web, iOS, and Android.” This announcement marked the conclusion of a phased deployment that had been unfolding over several weeks.
The feature, which first began appearing for select users in mid-May ahead of Google I/O 2026, allows users to control how much reasoning effort the AI model applies to their queries. This capability addresses a practical reality of working with advanced AI systems, namely that different tasks call for different levels of deliberation. Some queries demand careful, extended reasoning to produce thoughtful and accurate results, while others can be handled efficiently with a more streamlined approach. By placing this choice in the hands of users, Google acknowledges the diverse needs of its audience and provides a tool for optimizing the balance between speed and depth of analysis.
In terms of how the feature functions, the Thinking Levels option appears at the bottom of the model picker in the Gemini app when users select either the Fast model, known as Gemini 3.5 Flash, or the Gemini 3.1 Pro model. Users can choose between two settings: “Standard,” which serves as the default option, and “Extended,” which directs the model to spend more time deliberating on complex topics before generating a response. This straightforward interface design makes the feature accessible to users of varying levels of technical sophistication, allowing them to adjust the AI’s reasoning behavior with minimal friction as they navigate their tasks.
There are, however, practical considerations associated with the more intensive reasoning mode. The Extended setting consumes more of a user’s usage quota, particularly when paired with the 3.1 Pro model, reflecting the greater computational resources required for deeper deliberation. Google’s support documentation indicates that free users operate under standard usage limits, while paid subscribers at the AI Plus, Pro, and Ultra tiers receive progressively higher allocations. This structure illustrates how Google has integrated the feature into its broader subscription framework, balancing the desire to make reasoning capabilities widely available with the economic realities of providing computationally demanding services at scale.
The timing of the completed rollout is notable, as it arrives roughly two weeks after Google introduced Gemini 3.5 Flash at its I/O 2026 developer conference on May 19. That model has attracted considerable attention, with Google stating that it outperforms the company’s previous flagship Gemini 3.1 Pro on most benchmarks while running four times faster than competing frontier models. Following its debut, Gemini 3.5 Flash became the default model in both the Gemini app and AI Mode in Search globally, underscoring Google’s confidence in the model’s capabilities and its central role in the company’s AI strategy. The integration of Thinking Levels with this new flagship model enhances the overall flexibility and power available to users.
Beyond the freely available Thinking Levels feature, Google maintains a more advanced reasoning capability reserved for its premium subscribers. Deep Think, the company’s most sophisticated reasoning mode, remains exclusive to Google AI Ultra subscribers. Built on top of Gemini 3.1 Pro, Deep Think employs parallel thinking techniques to tackle challenging problems in mathematics, science, and engineering that test even state-of-the-art models. This capability represents the upper tier of Google’s reasoning offerings, designed for users who require the most powerful analytical tools available within the Gemini ecosystem for demanding intellectual tasks.
The distinction between the freely available Thinking Levels and the subscriber-only Deep Think mode reflects Google’s carefully considered tiered approach to monetizing its AI capabilities. By offering meaningful reasoning controls to its broader user base while reserving its most advanced capabilities for paying subscribers, Google seeks to expand access to useful features while preserving incentives for users to upgrade to premium tiers. This balanced strategy allows the company to serve a wide spectrum of users, from those seeking everyday assistance to professionals and researchers requiring the most capable tools available. As Google continues to refine and expand its Gemini offerings, the completed rollout of Thinking Levels stands as a clear example of how the company is working to make its AI both more powerful and more adaptable to the varied needs of its global audience, signaling an ongoing commitment to placing greater control in the hands of those who use its technology.
