BlockBeats News, May 20, Google DeepMind today officially released the new generation model series Gemini 3.5, and first launched Gemini 3.5 Flash. The official stated that the model combines "cutting-edge intelligence" with "actionability," focusing on AI agents and programming scenarios, and can handle complex, long-term tasks.
Gemini 3.5 Flash is now open to global users and can be accessed through the Gemini App, Google Search AI Mode, Google AI Studio, Android Studio, and the enterprise platform Gemini Enterprise. Google also revealed that Gemini 3.5 Pro has been internally deployed and is expected to be officially launched next month.
According to the data released by Google, Gemini 3.5 Flash outperformed Gemini 3.1 Pro in intelligent agent and programming benchmarks such as Terminal-Bench 2.1, GDPval-AA, MCP Atlas, and achieved 84.2% in the multimodal understanding test CharXiv Reasoning. Google stated that its output speed is four times faster than other cutting-edge models.
Google emphasized that the core capability of 3.5 Flash lies in "long-term intelligent agent tasks," including automated planning, multi-step execution, tool invocation, and deploying multiple collaborative sub-agents. The official stated that development, auditing, or financial processes that used to take several days or even weeks can now be completed in a shorter time, usually at less than half the cost of other cutting-edge models.
In terms of business implementation, companies such as Shopify, Salesforce, Macquarie Bank, Ramp, Databricks, and Xero have started testing or deploying Gemini 3.5 Flash.
Google also unveiled the new personal AI agent "Gemini Spark" built on Gemini 3.5 Flash. The product can assist users with digital transactions around the clock and actively perform tasks with user authorization. Currently, Gemini Spark is open to some beta testers and plans to launch a Beta version next week for U.S. Google AI Ultra subscribers.
Furthermore, Gemini 3.5 Flash has now become the default model for the Gemini App and Google Search AI Mode, supporting the generation of more dynamic and interactive Web UI interfaces.
