Google rolled out the Gemini 3.5 family alongside a new video-generation model, Omni, at its annual Google I/O 2026 on 19 May, positioning its AI strategy around lightweight models and autonomous agents as competition intensifies with OpenAI and Anthropic ahead of their expected public market debuts later in 2026.

The announcements focused more on practical deployment than benchmark dominance, reflecting Google’s effort to defend its search leadership whilst expanding AI capabilities across products already serving billions of users.

Gemini 3.5 architecture

Google launched Gemini 3.5 Flash immediately, introducing a lighter-weight model designed for coding, automation, and agent-based workflows whilst maintaining fast response times.

The model now powers the Gemini app globally and AI Mode in Google Search, forming the foundation for several new features, including the Gemini Spark agent announced during the conference.

Gemini 3.5 Pro, the heavier-weight version currently undergoing internal testing, will launch publicly next month, targeting enterprise use cases requiring stronger reasoning capabilities.

CEO Sundar Pichai said the Gemini 3.5 family was “designed to power more capable AI agents that can manage complex and longer tasks more effectively.”

Omni: Google’s AI “world model”

Google also introduced Gemini Omni, described as a “world model” capable of simulating physical reality through video generation and editing.

The system combines text prompts, images, and video clips to generate cinematic-quality outputs whilst understanding physics, historical references, scientific logic, and cultural context. Google positioned Omni as a system designed to create coherent narratives rather than simply visually realistic footage.

“Take a video you shot and just ask Omni to change what’s happening,” Google explained. “The AI can edit the action, add in new characters or objects.”

Gemini Omni Flash became available immediately for Google AI Plus, Pro, and Ultra subscribers through the Gemini app and Google Flow platform. Integration with YouTube Shorts and the YouTube Create app launches this week at no additional cost.

Gemini Spark agent

The company also unveiled Gemini Spark, a general-purpose AI agent embedded within the Gemini app.

Spark reasons across connected applications and performs tasks on the users’ behalf whilst operating under user direction. The system automates workflows and continues running tasks in the background, placing Google directly into competition with Anthropic’s Claude agents and OpenAI’s operator tools.

Spark enters beta next week for trusted testers and Google AI Ultra subscribers.

Google simultaneously announced Android Halo, a feature displaying real-time updates showing what AI agents are doing directly from the top of users’ phone screens without interrupting ongoing activity.

Google’s competitive positioning

The announcements arrive as investors increasingly scrutinise Google’s AI spending, whilst OpenAI and Anthropic continue attracting market attention ahead of their anticipated IPOs.

Google faces dual pressure: maintaining traditional search revenue as users migrate toward conversational AI tools, whilst proving its AI investments create defensible competitive advantages rather than simply matching rivals’ features.

The emphasis on lightweight models like Flash and autonomous systems like Spark suggests Google prioritises large-scale deployment and ecosystem integration over raw benchmark performance.

By integrating Omni directly into YouTube, a platform serving over 2 billion monthly users, Google also demonstrated a strategy extending AI monetisation beyond search into creator tools and video production infrastructure.

Implications for marketers and creators

For marketers, advertisers, and content creators, Omni’s video-generation capability could significantly alter production economics.

If AI systems can generate cinematic-quality video from text prompts and simple edits, production timelines and creative costs could compress dramatically whilst increasing iteration speed for campaigns and digital content.

Whether AI-generated quality can consistently replace professional human production remains debated. However, the technology lowers entry barriers for smaller creators and brands previously excluded from high-end video production because of cost and resource constraints.

ALSO WATCH:MARKETING EDGE ONTV