Google unveils Gemini 3.5 Flash

Google used its I/O conference on May 19 to introduce Gemini 3.5 Flash, a new model the company said is built for coding and AI agents and is the first release in its Gemini 3.5 family. In separate product updates last week, OpenAI introduced gpt-realtime-2 for voice applications and began rolling out a personal-finance feature in ChatGPT that can connect to users’ financial accounts. Together, the launches show the largest AI companies pushing on different parts of the market at once: faster agent models, more capable speech systems and consumer products tied to sensitive personal data. European officials are also watching how companies disclose advanced systems to regulators, according to recent reports. ### What exactly did Google launch at I/O? Google said Gemini 3.5 Flash is the first model in a new Gemini 3.5 series and described it as combining “frontier intelligence with action.” In its I/O materials, the company positioned the model for real-world agent workflows, with an emphasis on speed, coding and tool use. (blog.google) At Google I/O 2026, the company also said Gemini 3.5 Flash outperforms Gemini 3.1 Pro across almost all benchmarks while running four times faster than other frontier models. Google tied that claim to its broader push beyond prompt-based assistants toward systems that can take actions through tools and developer platforms. (blog.google) ### Why is OpenAI talking about voice at the same time? OpenAI published documentation for gpt-realtime-2 last week and said the model supports configurable reasoning effort, with higher reasoning settings increasing latency and output token usage. The company’s developer site lists pricing at $4 per 1 million text input tokens and $24 per 1 million text output tokens, with separate audio pricing. (blog.google) On May 7, OpenAI said it was introducing a new generation of realtime voice models in the API that can “reason, translate, and transcribe as people speak.” That places OpenAI’s latest push in low-latency speech systems rather than the coding-and-agent position Google emphasized for Gemini 3.5 Flash. (developers.openai.com) ### What is the new ChatGPT finance feature? OpenAI said on May 15 that it was releasing a preview of a personal-finance experience in ChatGPT to Pro users in the United States. The feature lets users connect financial accounts, view a dashboard showing where money is going, and ask questions grounded in their own financial context, goals and priorities. (openai.com) OpenAI’s help documentation says Finances is currently available to Pro users in the United States on web and iOS and is rolling out gradually. The company said users can connect supported accounts and view spending, bills, subscriptions, net worth and investment information in one place. ### Why are regulators part of this story too? (openai.com) A report published on May 20 by El Periódico, cited in secondary coverage, said European officials and the financial sector have been unsettled by recent AI claims and are paying attention to access and disclosure around advanced systems. The same reporting said OpenAI has shared its GPT-5.5-Cyber model with EU authorities, while Anthropic has faced questions over access in Europe. (help.openai.com) OpenAI said earlier this month that GPT-5.5-Cyber is available in limited preview through its Trusted Access for Cyber program. Reports about EU access suggest that regulatory cooperation itself is becoming part of how advanced AI products are introduced, though that characterization comes from the reporting rather than from a formal EU policy document in the materials reviewed here. (techrepublic.com) ### What comes next from here? Google’s I/O materials say Gemini 3.5 Flash is the opening release in the Gemini 3.5 line, indicating more models in that family are likely to follow. OpenAI’s finance product is still labeled a preview and is being rolled out gradually to U.S. Pro users, while its realtime voice lineup is already available through the API for developers building speech applications. (siliconangle.com) (blog.google)