Google’s Gemini 3.5 Flash: Pioneering the Future of Autonomous AI Agents
At the recent Google I/O developer conference, Google unveiled Gemini 3.5 Flash, its most advanced AI model to date, designed to revolutionize coding and autonomous AI agents. This model can independently execute coding pipelines, manage research projects, and, in internal tests, has even built an operating system from scratch.
This launch marks a significant shift in Google’s AI strategy, moving from conversational tools to agentic tools capable of planning, building, and iterating on tasks with minimal human intervention. Koray Kavukcuoglu, DeepMind’s chief technologist, highlighted that Gemini 3.5 Flash offers an exceptional combination of quality and low latency, outperforming the previous frontier model, 3.1 Pro, in coding, agentic tasks, and multimodal reasoning. Notably, it operates four times faster than other frontier models, with an optimized version achieving speeds up to twelve times faster without compromising quality.
The model’s speed is crucial for agentic work, enabling multiple AI agents to run concurrently on long-term tasks. During the I/O conference, Google engineer Varun Mohan demonstrated agents working on separate components and then integrating them to build a complete operating system within Antigravity, Google’s agentic development platform and integrated development environment (IDE). Kavukcuoglu emphasized that Gemini 3.5 Flash was co-developed with Antigravity to provide a native environment for agents to operate effectively.
Beyond demonstrations, Gemini 3.5 Flash’s agentic capabilities are already making a significant impact among partners. Banks and fintech companies are automating multi-week workflows, and data science teams are uncovering insights in complex data environments. The model can run autonomously for several hours, though it will pause and seek user input when encountering decision points or permission issues requiring human judgment.
Looking ahead, Google plans to release the 3.5 Pro model, designed to work in tandem with 3.5 Flash. Tulsee Doshi, Google’s senior director and head of product, explained that 3.5 Pro will serve as the orchestrator and planner, leveraging Flash as various sub-agents. This collaboration allows for a balance between reasoning power and efficient tool use capabilities.
Gemini 3.5 Flash is now the default model in the Gemini app and AI Mode in Search globally. Google also announced agentic capabilities coming to Search, enabling users to create, customize, and manage AI agents directly on the platform. Additionally, the new model will power Gemini Spark, Google’s personal AI agent designed to operate 24/7 to assist consumers in managing their digital lives.
As Google introduces this level of AI capability to the general public, it faces increased scrutiny. The company is currently dealing with a lawsuit following a tragic incident involving a man who nearly committed a mass casualty event and died by suicide after weeks of interacting with Gemini last year. The potential for harm increases with the broader availability of powerful autonomous agents. In response, Google has strengthened cyber and Chemical, Biological, Radiological, and Nuclear (CBRN) safeguards in Gemini 3.5 and improved its calibration to engage with sensitive questions thoughtfully rather than refusing them outright.
Gemini 3.5 Flash is now generally available through Antigravity, the Gemini API, and Gemini Enterprise, as well as via the Gemini app and AI mode in Search.
