Agentic AI techniques have gotten central to Google’s long-term AI technique.
Google has announced Gemini 3.5, a brand new household of AI fashions designed to mix frontier-level reasoning with stronger agentic capabilities throughout shopper, developer and enterprise merchandise.
The corporate is launching the sequence with Gemini 3.5 Flash, which it describes as its strongest agentic and coding mannequin to this point. Google stated the mannequin is constructed for advanced, long-horizon duties, together with multi-step workflows, coding, doc evaluation and enterprise automation.
In response to Google, Gemini 3.5 Flash outperforms Gemini 3.1 Professional on a number of coding and agentic benchmarks, together with Terminal-Bench 2.1, GDPval-AA and MCP Atlas, whereas additionally bettering multimodal reasoning. The corporate additionally claimed the mannequin is considerably quicker than different frontier fashions when measured by output tokens per second.
Google stated the mannequin can help agentic workflows via its Antigravity growth platform, together with using collaborative subagents for coding, monetary doc preparation, knowledge evaluation and different advanced enterprise duties. It additionally highlighted richer multimodal capabilities, together with interactive internet interfaces, visible reasoning and generative UI experiences.
Gemini 3.5 Flash is accessible via the Gemini app, AI Mode in Google Search, the Gemini API in Google AI Studio, Android Studio, Google Antigravity and Gemini Enterprise merchandise. Google additionally stated Gemini 3.5 Professional is getting used internally and is anticipated to roll out subsequent month.
The announcement additionally launched Gemini Spark, a private AI agent powered by Gemini 3.5 Flash. Google stated Spark is designed to run constantly, serving to customers navigate digital duties and take motion underneath their route. It’s beginning with trusted testers, with a beta deliberate for Google AI Extremely subscribers within the US.
Alongside efficiency enhancements, Google stated Gemini 3.5 was developed underneath its Frontier Security Framework. The corporate stated it strengthened safeguards associated to cybersecurity and chemical, organic, radiological and nuclear dangers, whereas including security coaching and interpretability instruments meant to cut back dangerous outputs and mistaken refusals.
The launch displays a wider trade shift from conversational AI assistants in direction of techniques that may plan, coordinate and execute duties throughout digital environments, with main AI firms more and more competing on agentic workflows, coding efficiency, multimodal interplay and enterprise integration.
Why does it matter?
Gemini 3.5 exhibits how the AI race is transferring past chatbot efficiency in direction of techniques that may act throughout software program, workflows and enterprise environments. Sooner agentic fashions may assist automate coding, evaluation and enterprise operations, however additionally they increase governance questions round supervision, security, accountability and the way a lot autonomy customers and organisations ought to give to AI brokers.
Would you prefer to be taught extra about AI, tech and digital diplomacy? In that case, ask our Diplo chatbot!