Google introduces Gemini 3 Flash as default AI model for the Gemini app

Following the release of Gemini 3 Pro and Gemini 3 Deep Think, Google, the Mountain View tech giant, has announced Gemini 3 Flash, its new lightweight AI model. This AI model will run as a default model in the Gemini app and AI mode in search, replacing the 2.5 Flash model. It is also said to outperform the 2.5 Flash model and Gemini 2.5 Pro in terms of performance and reasoning.

Gemini 3 Flash: Faster, smarter, and more efficient AI

According to Google’s blog post, the Gemini 3 Flash scored 33.7% in Humanity’s Last Exam, whereas the Gemini 2.5 Flash scored 11%. The new AI model also scored 81.2% at the MMMU-Pro test, which determines AI understanding and reasoning using text, images, and other inputs.

With the new AI model, Google is said to have retained the core intelligence and reasoning abilities of Gemini 3.0, while also making it fast and efficient like a lightweight AI model. With Gemini 3 Flash’s multimodal reasoning capabilities, it can efficiently analyse visuals and sounds, helping users get answers faster. Google provided an example saying, “You can ask Gemini to understand your videos and images and turn that content into a helpful and actionable plan in just a few seconds.”

In addition, users can also speak their idea out loud in layman’s language, and it can transform the thoughts into a functioning app within minutes. The new Gemini model has started to roll out to everyone in the Gemini app and AI Mode in Search. It's also available for preview in Gemini API in Google AI Studio, Google Antigravity, Vertex AI, and Gemini Enterprise.