Google launches cutting-edge intelligent model Gemini 3 Flash

According to the official Google blog, Google has launched Gemini 3 Flash, positioned as a high-speed, low-cost cutting-edge intelligent model, with API access now open to Vertex AI. The official statement claims it achieves 90.4% on GPQA Diamond and 33.7% on Humanity's Last Exam without tools, being faster and cheaper than 2.5 Pro, even outperforming the old version at the lowest thinking level.

The pricing is $0.50 per million tokens for input and $3 per million tokens for output (audio input is $1 per million), supporting context caching (up to 90% cost reduction) and Batch API (50% cost reduction, higher speed). New features include stronger visual/spatial reasoning and code execution, which can be used for scaling, counting, and editing images, and have been integrated into Google AI Studio, Antigravity, Gemini CLI, Android Studio, and Vertex AI.