Today, we're expanding the Gemini 3 family with the release of the Gemini 3 Flash, which offers pioneering intelligence built for speed at a fraction of the cost. In this release, we'll make the next-generation intelligence of Gemini 3 available to anyone using Google services.
Last month we started work on Gemini 3 Gemini 3 Pro AND Gemini 3 Think deeply mode, and the response was amazing. Since the day of launch, we have been processing over 1T tokens per day in our API. We've seen you use Gemini 3 to vibration code simulations to explore complex topics, build and design interactive games and understand all types multimodal content.
With Gemini 3, we pioneered performance based on complex reasoning, multimodal and vision understanding and agentic and vibratory coding tasks. Gemini 3 Flash maintains this foundation, combining Pro-class reasoning of Gemini 3 with Flash-level latency, performance and cost. Not only does it enable you to perform everyday tasks with improved reasoning, but it is also our most impressive agentic workflow model yet.
Starting today, Gemini 3 Flash will be made available to millions of people around the world:
