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

In a groundbreaking move, Google has introduced Gemini 3 Flash, a cutting-edge lightweight AI model that will serve as the default for the Gemini app and the AI mode in search functionalities. This new model supersedes the previous 2.5 Flash version and promises enhanced performance and reasoning capabilities. According to a recent post from Google, Gemini 3 Flash has achieved impressive scores, including 33.7% in the Humanity’s Last Exam, a significant improvement over the 11% scored by its predecessor, the Gemini 2.5 Flash. Furthermore, it excelled in the MMMU-Pro test, achieving 81.2%, which evaluates an AI’s comprehension and reasoning through various inputs such as text and images. Gemini 3 Flash combines the advanced intelligence and reasoning skills of the Gemini 3.0 model with the agility and efficiency typical of lightweight AI systems. Notably, its multimodal reasoning capabilities allow it to swiftly analyze visuals and sounds, enabling users to receive answers in record time. Google illustrated this feature by stating that users can ask Gemini to interpret their videos and images, transforming that content into actionable plans within seconds. Moreover, the model offers a unique capability where users can verbally express their ideas in simple terms, and Gemini 3 Flash can convert those thoughts into a functional app in mere minutes. This innovative model is currently being rolled out to all users via the Gemini app and AI Mode in Search, and it is also accessible for preview in the Gemini API through Google AI Studio, Google Antigravity, Vertex AI, and Gemini Enterprise.