Google I/O: Gemini 4 buzz

Google’s next I/O is still five weeks away, but the chatter around “Gemini 4” is really a reaction to Google’s March and April 2026 AI rollouts and to what the company has already scheduled for May 19-20. (blog.google) Google said on February 17 that I/O 2026 will run May 19-20 at Shoreline Amphitheatre in Mountain View, California, with updates planned across Gemini, Android, Chrome and Cloud. The official preview also promised “the latest Gemini model updates,” but it did not name a “Gemini 4” release. (blog.google) The video side of the buzz is easier to pin down. Google DeepMind’s Veo page now centers on Veo 3.1, which generates video with native audio, including sound effects, ambient noise and dialogue, and Google’s developer changelog says a lower-cost Veo 3.1 Lite Preview launched on March 31, 2026. (deepmind.google, ai.google.dev) That matters because Google is no longer treating these models as lab demos. By March 2026, Google was expanding “Personal Intelligence” across Search’s Artificial Intelligence Mode, the Gemini app and Gemini in Chrome in the United States, tying model features to products that already have mass audiences. (blog.google) Google used the same playbook at I/O 2025. It said Artificial Intelligence Overviews had reached 1.5 billion monthly users in 200 countries and territories, rolled Artificial Intelligence Mode out in U.S. Search, and brought Veo 3 and Imagen 4 into the Gemini app. (blog.google, blog.google) For creators, the practical shift is that Google’s video tools are attached to consumer and developer channels at the same time. The Veo page offers “Try in Gemini,” “Try in Flow” and “Build with Veo,” while the Gemini app update from May 2025 said Veo 3 was built directly into the app and Vertex Artificial Intelligence access was available for builders. (deepmind.google, blog.google) For developers, Google has also been filling in the plumbing around those models. The Gemini application programming interface changelog lists multimodal embeddings for text, image, video, audio and PDF inputs on March 10, 2026, audio-to-audio Gemini 3.1 Flash Live on March 26, and Veo 3.1 Lite on March 31. (ai.google.dev) The naming confusion comes from Google’s broader model lineup. In April 2026, Google launched Gemma 4, its open model family, while its public Gemini news pages and release notes were still centered on Gemini 3.1, not a formally announced Gemini 4. (blog.google, blog.google, ai.google.dev) So the safest read on the “Gemini 4 buzz” is that people are reacting to real momentum in Google’s multimodal stack, especially video and voice, before the company has actually put a Gemini 4 label onstage. The next hard date is May 19, when I/O 2026 starts and Google says those Gemini updates will be shown live. (developers.googleblog.com)