Introducing 'Gemini 3.5,' a cutting-edge intelligence with execution capabilities
https://blog.google/intl/ko-kr/company-news/technology/gemini-3-5-kr/
Part of the content is quoted.
Today, Google introduces 'Gemini 3.5 (Gemini 3.5),' the latest product line combining cutting-edge intelligence with execution capabilities. This marks a significant leap forward in building more capable and intelligent agents. As the first installment in the series, we are unveiling '3.5 Flash.' This model delivers consistent state-of-the-art performance in agent and coding domains, excelling at complex and long-horizon tasks that are useful in real life.
Starting today, billions of users worldwide will have access to 3.5 Flash.
For General Users: Available through the Gemini app and AI mode in Google Search.
For Developers: Provided through Google Antigravity (Google's agent development platform), Google AI Studio, and the Gemini API within Android Studio.
For Enterprise Customers: Accessible through the Gemini Enterprise Agent Platform and Gemini Enterprise.
Furthermore, Google is accelerating the development of 3.5 Pro, which is already being used internally and scheduled for official release next month.
3.5 Flash (3.5 Flash): Stable Performance Support in Agent Functionality and Coding Domains
Gemini 3.5 Flash maintains the rapid processing speed characteristic of the Flash series while offering exceptional intelligence comparable to large flagship models in various aspects. It is the most powerful agent and coding model released so far, surpassing the performance of Gemini 3.1 Pro. In demanding coding and agent benchmarks such as Terminal-Bench 2.1 (76.2%), GDPval-AA (1656 Elo), and MCP Atlas (83.6%), 3.5 Flash outperformed 3.1 Pro, and it is also leading in multimodal understanding (84.2% according to CharXiv inference standards). In terms of tokens per second output, Gemini 3.5 Flash offers a speed four times faster than other frontier models.
[The rest of the content is omitted]
