Google expanded its artificial intelligence ambitions at Google I/O 2026 with the launch of Gemini 3.5 Flash, a faster AI model designed for coding and autonomous task execution. The company also introduced Gemini Spark, a personal AI agent, and Gemini Omni, a multimodal video generation model capable of producing polished videos from different forms of input.

The announcements reflect Google’s broader push to turn Gemini into an ecosystem of AI products spanning consumer services, enterprise software, and developer tools. The company increasingly focuses on AI systems that can complete actions and workflows rather than simply respond to prompts.

Gemini 3.5 Flash will roll out immediately across the Gemini app and AI Mode in Google Search, while Gemini 3.5 Pro remains in internal testing ahead of a planned launch next month.

Faster coding performance

Google positioned Gemini 3.5 Flash as its most advanced Flash model yet, with improvements aimed at coding, reasoning, and long-horizon task execution. Unlike earlier lightweight AI models that often traded capability for speed, the new release attempts to deliver both.

According to benchmark data shared by the company, Gemini 3.5 Flash surpassed Gemini 3.1 Pro on several coding and agentic evaluations, including Terminal-Bench 2.1, GDPval-AA, and MCP Atlas. The model also posted strong multimodal reasoning performance on the CharXiv benchmark.

The company claims Gemini 3.5 Flash produces outputs four times faster than competing frontier AI models while maintaining high reasoning performance. That balance, Google argues, makes the model suitable for complex tasks such as application development, maintaining large software projects, and preparing financial documents.

Google also emphasized the model’s ability to handle “long-horizon agentic tasks,” a growing category of AI systems designed to execute multi-step objectives with limited human supervision.

Push into AI agents

Much of Google’s presentation centered on AI agents, which have become a major battleground in the broader AI industry. On the Gemini 3.5 Flash product page, Google described the model family as combining “frontier intelligence with action” and called it “a major leap forward in building more capable, intelligent agents.”

The company paired the new model with an updated version of Google Antigravity, its agent-focused development platform. The platform allows developers to deploy collaborative subagents capable of executing workflows and coding tasks simultaneously.

Google said the system can sustain high performance while managing complex multi-step operations under supervision. The company also highlighted the model’s ability to support richer web interfaces and more interactive graphics generation.

Developers can access Gemini 3.5 Flash through Google AI Studio, Android Studio, the Gemini API, and Antigravity. Enterprise customers will gain access through Gemini Enterprise and Gemini Enterprise Agent Platform.

Expanding the Gemini lineup

Beyond Gemini 3.5 Flash, Google used the event to introduce new AI products aimed at broadening Gemini’s role across productivity and media generation.

Gemini Spark is designed as a personal AI agent that can assist users with tasks and workflows across Google services. Meanwhile, Gemini Omni focuses on multimodal video creation, allowing users to generate refined videos from different forms of input.

Google did not provide detailed technical specifications for Spark or Omni during the keynote. Still, the launches reinforce the company’s aggressive effort to compete with rivals racing to build faster, more capable generative AI systems for both consumers and businesses.