Google Bets Big on AI Agents with Gemini 3.5 Flash

Google has unveiled Gemini 3.5 Flash, a cutting-edge AI model that the company claims is its strongest yet for coding and autonomous AI agents. Introduced at the annual I/O developer conference, Gemini 3.5 Flash can independently execute coding pipelines, manage research projects, and even build an operating system from scratch, according to internal tests. The release signals a significant shift in Google's approach to AI, from focusing on conversational tools to developing AI as an agentic tool.

This new approach enables AI to plan, build, and iterate on real-world tasks with minimal human input, rather than simply answering questions. Koray Kavukcuoglu, DeepMind's chief technologist, highlighted the model's capabilities, stating that '3.5 Flash offers an incredible combination of quality and low latency.' He noted that it outperforms Google's latest frontier model, 3.1 Pro, on nearly all benchmarks, including coding, agentic tasks, and multimodal reasoning. Moreover, Gemini 3.5 Flash is four times faster than other frontier models, with an optimized version that is 12 times faster while maintaining the same quality.

The speed of Gemini 3.5 Flash is central to its design for agentic work, where multiple AI agents run simultaneously on long-running tasks. Google engineer Varun Mohan demonstrated this capability on stage at I/O, showcasing agents spawning off to work on separate components before coming together to build a full operating system within Antigravity, the company's agentic development platform and IDE. According to Kavukcuoglu, Flash 3.5 was co-developed with Antigravity to provide a native environment for agents to execute and work.

Google says the benefits of Gemini 3.5 Flash are already being realized among its partners, such as banks and fintechs automating multi-week workflows, or data science teams uncovering insights in complex data environments. The model can run autonomously for multiple hours, although it may pause and request user input when encountering decision points or permission issues that require human judgment. As Google prepares to release its forthcoming 3.5 Pro model, the two are designed to work in tandem, with 3.5 Pro serving as an orchestrator and planner that leverages Flash as a sub-agent.

The company is also integrating agentic capabilities into Search, allowing users to create, customize, and manage AI agents directly on the platform. Gemini 3.5 Flash is now available globally as the default model in the Gemini app and AI Mode in Search, and will power Gemini Spark, a new personal AI agent designed to assist consumers in managing their digital lives 24/7.