Top five announcements from Google I/O 2025
AI takes centre stage, with powerful tools for creators, coders, and everyday users

Google I/O 2025 has wrapped up at the Shoreline Amphitheatre in Mountain View, and this year's event was nothing short of a technological leap.
From bold advances in generative AI to new tools for developers, the company showcased its vision for the future across software, devices, and digital experiences. Here are the five standout announcements that stole the show.
Gemini Ultra
Priced at $249.99 per month, Gemini Ultra is not for everyone – but it is a peek into where AI is heading. With access to Deep Think in Gemini 2.5 Pro, a new video generator (Veo 3), an editing app (Flow), and tools like NotebookLM and Whisk, this elite tier is built for creators and power users. Subscribers also get 30TB of cloud storage and agentic AI tools powered by Project Mariner.
Veo 3
Veo 3, Google's latest video-generation AI model, can now generate not just visuals but audio – including background noise and speech. The upgrade means higher-quality clips and greater realism, with integration into Google's Gemini chatbot for AI Ultra users. It is a notable step forward in automated content creation.
Imagen 4
Google's image generation model, Imagen 4, impressed with its speed and detail. Able to render fabric textures and animal fur with photorealistic precision, it supports 2K resolution and multiple styles. Google plans to release a version up to 10 times faster, pushing real-time design even closer to reality.
Project Astra
Project Astra is Google's most advanced AI assistant yet – fast, context-aware, and capable of seeing, understanding, and responding in real time. Demonstrated with a smartphone camera, Astra could identify objects, remember locations, and answer complex questions on the fly. Whether helping someone find their glasses or explaining how a gadget works, it responded near-instantly with a natural voice.
Project Mariner
Perhaps the most futuristic demo came from Project Mariner. This agentic AI can browse websites, buy groceries, or book tickets – all by understanding your instructions. It marks a shift from assistant to autonomous agent, one that could redefine how we interact with the web.