Google has launched Gemini 3 Flash as its new default AI model across the Gemini app and Search, saying it delivers faster performance and stronger reasoning skills than previous releases. The company claims Gemini 3 Flash not only surpasses 2.5 Flash, but also comes close to matching its advanced Pro model. The move strengthens Google’s position in the competitive AI market, where OpenAI recently released GPT-5.2. With Gemini 3 Flash now becoming the primary experience for millions of users, Google appears set to accelerate adoption and challenge competitors directly.

Gemini 3 Flash sits just below the Pro tier but introduces a clear leap in output quality. One of the major performance highlights is a score of 33.7 per cent on the Humanity’s Last Exam test, which evaluates broad reasoning. This is a sharp rise from Gemini 2.5 Flash at 11 per cent, and close to both Gemini 3 Pro at 37.5 per cent and GPT-5.2 at 34.5 per cent. On the MMMU-Pro benchmark, which measures multimodal and reasoning ability, the model scored 81.2 per cent, outperforming leading AI models. These results suggest stronger interpretation, accuracy, and problem-solving skills, even without tool assistance.

Users around the world will now see Gemini 3 Flash as the standard model in the Gemini app and Search. It can interpret text, images, audio and video, providing clearer and more context-aware responses. Users can still manually switch to Pro for high-level coding, analysis and mathematical tasks. Flash is designed for everyday speed, offering rapid output while processing media-rich prompts. It also supports visual elements like tables, images and charts in responses, expanding flexibility for creative and educational uses.

Google says enterprise users are already adopting Gemini 3 Flash through Vertex AI and Gemini Enterprise with partners including JetBrains, Figma, Cursor, Harvey and Latitude. Flash is optimised for visual search, video review and data scanning, making it suitable for business tasks that rely on fast turnaround. Pricing is set at $0.50 per million input tokens and $3.00 per million output tokens, slightly higher than the previous Flash model but potentially more efficient due to reduced token usage.

Google also confirmed wider access to Gemini 3 Pro in the US and extended availability of the Nano Banana Pro image model. These changes aim to build a full ecosystem that works for casual users, developers and advanced workflows. Competition in the AI space continues to intensify, with OpenAI pushing new releases and usage growth, though Google’s latest update signals a strong counter. For now, the rollout of Gemini 3 Flash places Google at a new speed and capability threshold, shaping how millions will interact with AI by default.