Google unveiled Gemini 3.5 Flash on Tuesday, a new artificial intelligence model that the company asserts is its most potent offering to date for coding and autonomous AI agents. Introduced at Google's annual I/O developer conference, this model demonstrates capabilities such as independently executing coding pipelines, managing intricate research projects, and even, in internal testing environments, constructing an entire operating system from the ground up.
This release signifies a strategic pivot for Google, moving beyond the perception of AI primarily as a conversational tool to establishing it as a powerful agentic one. The focus is shifting from merely answering queries to enabling AI to plan, build, and iterate on complex tasks with minimal human intervention.
"3.5 Flash offers an incredible combination of quality and low latency," stated Koray Kavukcuoglu, DeepMind’s chief technologist, to reporters on Monday prior to the public launch. He further elaborated that "It outperforms our latest frontier model, 3.1 Pro, on nearly all the benchmarks," encompassing critical areas like coding, agentic tasks, and multimodal reasoning.
Kavukcuoglu also highlighted the model's impressive speed, noting it is four times faster than other frontier models—a velocity particularly beneficial for coding and agentic operations. Google, however, has "taken it to another level" by developing an optimized version of Flash that achieves a remarkable 12 times faster performance while maintaining the same high quality.
This exceptional speed is fundamental to Flash's design, especially for agentic workflows where multiple AI agents operate concurrently on long-duration tasks, as explained by Kavukcuoglu. During a demonstration at I/O, Google engineer Varun Mohan showcased how agents could independently branch off to develop distinct components before converging to assemble a complete operating system within Antigravity, Google’s dedicated agentic development platform and integrated development environment (IDE).
Kavukcuoglu confirmed that Flash 3.5 was specifically co-developed with Antigravity to provide agents with a "native environment where they can live, work, and execute."
Concurrent with these announcements at I/O, Google also launched Antigravity 2.0, a standalone desktop application meticulously crafted for agent-first development.
The benefits of these advancements are extending beyond mere demonstrations. Google reports that 3.5 Flash’s agentic capabilities are already generating significant impact among its partners, including financial institutions and fintech companies automating multi-week workflows, and data science teams extracting valuable insights from highly complex data environments.
While the model can function autonomously for several hours, Tulsee Doshi, Google’s senior director and head of product, clarified that it is designed to pause periodically and request user input when it encounters a critical decision point or a permission issue necessitating human judgment.
Looking ahead, Google's forthcoming 3.5 Pro model is engineered to operate in close conjunction with 3.5 Flash.
"3.5 Pro becomes your orchestrator, your planner, and then it actually can leverage Flash to be the various sub-agents," Doshi informed TechCrunch. She elaborated, "I think it really comes down to where do you really want that reasoning power, where you actually want that larger model that can really push on the reasoning side versus where do you have tasks that really do merit good brute force tool use capabilities?"
Gemini 3.5 Flash is now the default model within the Gemini app and AI Mode in Search across the globe. At I/O, Google also announced that agentic capabilities would soon integrate into Search, enabling users to create, customize, and manage AI agents directly on the platform. Furthermore, the new model will power Gemini Spark, Google’s innovative personal AI agent designed for 24/7 operation to assist consumers in managing their digital lives.
However, providing such sophisticated AI capabilities to average consumers also brings heightened scrutiny. Google is currently involved in a lawsuit following an incident last year where a man nearly caused a mass casualty event and subsequently died by suicide after weeks of engaging with Gemini.
The potential for harm intensifies as powerful autonomous agents become more widely accessible. In response, Google states that Gemini 3.5 incorporates strengthened cyber and CBRN (Chemical, Biological, Radiological, and Nuclear) safeguards and is more finely calibrated to engage with sensitive questions thoughtfully, rather than outright refusing them.
Gemini 3.5 Flash is broadly available starting today through Antigravity, the Gemini API, and Gemini Enterprise, as well as via the Gemini app and AI mode in Search.
The Editorial Staff at AIChief is a team of professional content writers with extensive experience in AI and marketing. Founded in 2025, AIChief has quickly grown into the largest free AI resource hub in the industry.