Google has started rolling out Gemini 3 Flash, a more efficient version of its latest AI model, nearly one month after launching Gemini 3 Pro in November. The company states that Gemini 3 Flash delivers similar pro-grade reasoning performance to the flagship model at a lower cost, suiting it for daily tasks.
Benchmarks indicate Gemini 3 Flash exceeds Google’s prior models, such as Gemini 2.5 Pro. It also competes closely with OpenAI’s GPT-5.2. In the Humanity’s Last Exam benchmark, without tools like web search, Gemini 3 Flash scored less than one percentage point below GPT-5.2.
Gemini 3 Flash outperformed GPT-5.2 in other tests. On the MMMU-Pro multimodal benchmark, which evaluates understanding and reasoning across modalities, it achieved 81.2 percent compared to GPT-5.2’s 79.5 percent.
Google is deploying Gemini 3 Flash as the default model in the Gemini App and AI Mode within Search. “That means all of our Gemini users globally will get access to the Gemini 3 experience at no cost, giving their everyday tasks a major upgrade,” the company said.
In AI Mode, users in the US can now access Nano Banana Pro, Google’s image generator, directly from the chatbot. They select “Thinking with 3 Pro” and then “Create Images Pro” from the model picker.
