Google Launches Gemini 3 Flash as New Default App Model
Google releases the Gemini 3 Flash model, making it the default choice in the Gemini app and AI search mode. The fast new model matches or beats frontier competitors on key industry benchmarks.
Google releases the fast and affordable Gemini 3 Flash model to rival recent offerings from OpenAI. The company makes this new model the default option in the Gemini app globally and in AI mode for search. This launch arrives just six months after the previous generation and brings significant performance upgrades to the table.
The new Flash model outperforms its predecessor by a wide margin and matches the performance of top-tier models like Gemini 3 Pro and GPT-5.2 on several measures. It scores 33.7% on Humanity’s Last Exam benchmark and achieves an industry-leading 81.2% on the MMMU-Pro multimodality and reasoning benchmark. The model excels at identifying multimodal content, allowing users to upload videos, sketches, or audio recordings for immediate analysis and feedback.
Enterprise clients like JetBrains, Figma, and Cursor already use Gemini 3 Flash through Vertex AI and Gemini Enterprise. Developers access the model via an API preview and Google's new Antigravity coding tool. Additionally, Google expands access to Gemini 3 Pro for all U.S. search users and rolls out the Nano Banana Pro image model to more people.