[AINews] Google I/O 2026: Gemini 3.5 Flash, Omni (NanoBanana for Video), Spark (background agents), and Antigravity 2.0

Google has launched the Gemini 3.5 Flash model, touted as its strongest coding model yet, featuring a 1M-token context and a 65k max output. The introduction of Gemini Omni combines reasoning with generative media, allowing for text, image, video, and audio inputs, showcasing Google's leading capabilities. These advancements highlight Google's strengthened position in the AI market, processing over 3.2 quadrillion tokens monthly, reflecting a significant 7x year-over-year increase.