Google brings Gemini’s ‘Extended’ thinking to everyone, no subscription needed

Affiliate links on Android Authority may earn us a commission. Learn more.

Gemini has been going through a lot of changes, whether that’s a new UI with spirited animation or a new theming system that matches Gemini’s colors to your Android phone’s wallpapers. On top of that, Google’s new Gemini 3.5 models bring efficiency-related upgrades, while its Spark agent is ready for you to offload your mundane tasks that can be worked on in the background. Along with these, Gemini is getting additional controls that let you change how deeply it thinks.

Right around Google I/O 2026, we started seeing the initial rollout for different “Thinking” levels in Gemini . These come in two options: Standard and Extended. The latter, as you would expect, lets Gemini think more deeply and for longer before responding to your queries, along with step-by-step reasoning. While at the time of our previous reporting, the feature was only available for a few of our teammates, it’s now available more widely.

Josh Woodward, VP of Gemini at Google, recently announced in an X post that Thinking levels are now available to users across multiple platforms. That includes the web interface, as well as the mobile interfaces on Android and iOS.

You can choose the Thinking level by heading over to the model settings in Gemini. The “Standard” and “Extended” thinking options are available across Gemini 3.5 Flash and Gemini 3.5 Flash-light, and are available to both free and paying users.

Meanwhile, Gemini 3.1 Pro also gets a “Deep think” option in addition to the two others, as per the video shared by Woodward, though it is limited to AI Ultra users only .

While extended thinking can be expected to offer more profound analyses and better responses (even though it’s not guaranteed), you should be mindful that it’s likely to burn through your usage limits faster . Google has been under fire recently for implementing a new credit system that uses tokens based on the complexity of the prompt, so you end up tokenmaxxing , even if you don’t intend to. Although the giant has been working to plug these gaps , you’re better off with the standard thinking setting for low-impact queries or searches that don’t require Gemini to think thoroughly before answering.

Thank you for being part of our community. Read our Comment Policy before posting.