At Google I/O 2026, Sundar Pichai highlighted the company's focus on AI agents, showcasing two new models: Gemini Omni and Gemini 3.5 Flash. These advancements aim to improve understanding and multimodality, with Gemini Omni enabling natural language creation and Gemini 3.5 Flash delivering execution of tasks as part of agentic workflows.
Gemini Omni allows users to combine images, audio, video, and text as input to generate high-quality videos, leveraging Gemini's real-world knowledge. The model is set to roll out in the Gemini app, Google Flow, and YouTube Shorts.
Gemini 3.5 Flash has unwavering focus on delivering execution of tasks as part of agentic workflows, combining frontier intelligence with action. It has achieved a significantly higher score of 76.2% in the Terminal-bench 2.1 agentic terminal coding benchmark.
The Gemini app is also gaining new powers with the introduction of the Gemini Spark agent, which transforms it from an assistant to an active partner that does real work on behalf of users. Spark is web-based and will integrate with other Google apps, allowing for recurring tasks and background actions even when devices are powered off.
Google has also announced new MCP connections with popular creative platform Canva, underlined by Spark. Further connections with Adobe, Xiaomi, Samsung, CapCut, OnePlus, and Dropbox are planned.
