Google ships agentic Gemini Spark and Gemini Omni video model at I/O 2026

At I/O 2026, Google leaned hard into agents and multimodality. Gemini Spark debuted as an agentic AI tool, accompanied by a redesigned Gemini interface, background agents, app integrations, and a native macOS desktop app — a notable platform expansion for a Google AI product onto Apple hardware. The Gemini Omni creative model accepts image, audio, video, and text input, extending Google's generative-media push.

The framing — leadership declaring "we are firmly in our agentic Gemini era" — became the story's flashpoint. Coverage from Glass Almanac and 9to5Google captured developer alarm over scope creep, with practitioners questioning whether Gemini is genuinely useful beyond polished demos. Google also touted smart-glasses partnerships with Warby Parker, Gentle Monster, and Samsung, signaling a hardware-distribution play to rival Meta's Ray-Bans.

Competitively, the I/O slate positions Google against OpenAI's agent ambitions and Anthropic's Claude Code, with Google betting that distribution across Search, Android, Workspace, and now glasses gives its agents reach rivals can't match. The macOS app is a direct shot at meeting users where they already work.

The skeptical undercurrent is real: a 9to5Google piece titled "what are you actually doing with Gemini" crystallized the gap between agentic marketing and daily utility that many developers feel. The open question is whether Spark and Omni convert into sticky workflows or join the pile of impressive-but-unused agent demos — a credibility test that matters as Google asks users to trust agents with more autonomy.