Users Will Soon Be Able to Control Gemini's Cognitive Load Themselves

May 18, 2026, 2:12 pm | Read time: 2 minutes

Google is reportedly working on a new setting for its AI chatbot Gemini. Users will soon be able to decide whether the AI should respond particularly quickly or more thoroughly.

Gemini to Differentiate Between Speed and Thoroughness

In everyday use, AI chatbots are often expected to respond as quickly as possible–such as for brief summaries, writing assistance, or simple questions. However, it becomes more challenging with complex tasks like math problems, programming questions, or extensive analyses.

This is precisely where Google is apparently focusing with a new feature for Gemini: According to consistent reports, initial users can set how much “thinking effort” the AI should exert for a response.

Competition Must Fix It 4 Reasons Why Apple Has Failed With Siri

Apple is likely to launch the new AI-powered Siri in iOS 27 initially as a beta version.

Bloomberg Report Suggests Testing Phase Apple’s AI Siri Likely to Launch in iOS 27 Only as a Beta

New Setting Directly in the Model Selection

The new feature currently appears only in certain Gemini models. So far, it affects “Gemini 3 Flash Fast” and “Gemini 3.1 Pro.” In the model selection, an additional menu item called “Thinking Effort” appears.

Currently, there are two levels to choose from: “Standard” and “Advanced.” According to Google, “Standard” is intended for most everyday questions, while “Advanced” is supposed to deliver better results for complex problem-solving.

Also of interest: Google announces the largest Android update in years

Competitors Already Using Similar Concepts

With the new feature, Google would not be alone. OpenAI already offers a comparable control of thinking effort with ChatGPT, such as in Thinking or Pro models. Other AI providers are pursuing similar approaches: Anthropic integrates extended or adaptive thinking with Claude, while Perplexity offers a special reasoning mode for complicated search queries.

The feature is apparently not yet available to all users. It is still unclear when Google will broadly release the adjustable thinking effort. There is also the possibility that the company may decide against a general release after the limited test.

This article is a machine translation of the original German version of TECHBOOK and has been reviewed for accuracy and quality by a native speaker. For feedback, please contact us at info@techbook.de.