Google Launches Gemini 3.5 Flash and AI Agent Tools

Google has officially launched Gemini 3.5 Flash, a new AI model designed to prioritize agentic workflows over simple question-answering [1]. The model is now the default engine powering both the Gemini app and AI Mode within Google Search [1].

Google claims that Gemini 3.5 Flash delivers four times the speed of comparable frontier models at less than half the cost [1]. Despite its "Flash" branding—previously reserved for cheaper, lighter alternatives—the company reports that the new model outperforms the older Gemini 3.1 Pro on coding and agentic benchmarks [1]. Specifically, the model achieved a 76.2% score on Terminal-bench 2.1 and an 83.6% on MCP Atlas [1].

The shift toward "agentic" AI marks a strategic pivot for Google [1]. Rather than focusing solely on information retrieval, the company is building models capable of planning and executing multi-step tasks, such as managing workflows or booking appointments with minimal user input [1]. To support this, Google introduced Gemini Spark, a personal AI agent that runs continuously to perform actions on a user’s behalf [1]. This tool is currently available to trusted testers, with a broader beta for US-based Google AI Ultra subscribers scheduled for next week [1].

Alongside the Flash release, Google announced Gemini Omni, a model capable of generating and editing high-quality video through conversational prompts [2]. The company also updated its subscription pricing, lowering the cost of its top-tier AI Ultra plan from $250 to $200 per month, while introducing a new $100-per-month tier for developers and creators [2].

These updates arrive as Google intensifies its competition with OpenAI and Anthropic [2]. Following the launch of Gemini 3.0 in late 2025, which helped drive a 126% increase in Google’s stock price, the company is now looking to solidify its lead by embedding these agentic capabilities directly into its core consumer products [2].

While Gemini 3.5 Flash is available globally today, the company has already confirmed that a more powerful Gemini 3.5 Pro model is currently in internal testing, with a rollout expected next month [1]. Whether these agentic features can successfully transition from controlled benchmarks to reliable daily utility remains the primary test for Google’s latest AI roadmap.