Gemini Omni and Gemini 3.5 Flash launch at I/O 2026 with video focus and aggressive pricing

Demis Hassabis used the I/O 2026 keynote to position video generation at the heart of Google's AI strategy, unveiling Gemini Omni and the faster Gemini Omni Flash variant. The model translates between text, audio, images and video — users can prompt outputs like a 'claymation explainer of protein folding' from combinations of media. Logan Kilpatrick framed it on X as 'our new model that can create anything from any input — starting with video,' rolling out across the Gemini App, Flow, YouTube, with API support coming soon.

Gemini 3.5 Flash, the speed-tier companion, arrived with aggressive pricing marketing. Simon Willison's notes flagged a wrinkle: it's 3x the price of Gemini 3 Flash, but Google plans to use it across many of its own products — a sign Google is willing to subsidize its own surface area while still undercutting rivals at the API layer. Google also 3x'd Antigravity rate limits across all tiers to let developers stress-test 3.5 Flash. Agentic features Gemini Spark and Universal Cart shipped alongside.

The rollout was not friction-free. Google Flow credits do not automatically unlock Gemini Omni Flash video generation, creating tier confusion for users expecting parity. CNET-style commentary labeled Omni a likely accelerator of 'AI slop,' and a separate Medium piece argued 'Gemini Omni is not the completion of AGI' some are framing it as. Meta was repeatedly named as the rival Google is trying to narrow the gap with — alongside OpenAI's video efforts and Anthropic's text dominance.

What to watch: how API pricing for Omni shakes out versus Sora-class competitors, whether Antigravity's developer reception holds after a separate HN thread (553 pts) called it 'a bait and switch,' and how quickly the credit-tier confusion gets cleaned up before consumer fatigue sets in.