Google swaps its AI approach for specialized models

Google has announced new AI models specialized in security and programming. Tools such as SynthID and Google Vids take center stage. The announcement was made in September 2026.

Google announced new AI advances in September 2026. The focus included cybersecurity, programming and integration with devices such as Chromebook and Googlebook. The Gemini 4 Argon model was developed for complex tasks in programming, cybersecurity and professional activities. It set new records in software development and cybersecurity. SynthID, a detector of AI-generated content for images and videos, includes invisible watermarks applied during content generation. Google made its AI video creation tool, called Google Vids, available for free to personal and Workspace accounts.

Google Vids uses the Gemini Omni 1.1 Flash model to generate videos in 1080p. Google will launch a satellite with TPUs in low Earth orbit as part of Project Suncatcher. Project Suncatcher will test the resilience of AI chips in extreme space conditions. 1080p resolution offers enough quality for use in online meetings and presentations. Gemini 4 Argon can help programmers identify vulnerabilities in code. But the model still has no confirmed release date in Brazil.

Explore our newest frontier model — Image: blog.google

Google has strengthened its presence in AI with Gemini 4 Argon, which expands software development and cybersecurity capabilities. SynthID, a tool for detecting AI-generated content, has gained global support and integration with brands such as OpenAI and NVIDIA. These advances continue a line of innovation that Google began with Gemini 3.8 and SynthID Bio, reinforcing its position in the market.

Skip the typing and get things done just by talking — Image: blog.google

The company also expanded Google Vids, integrating Gemini Omni 1.1 Flash technology, which improves content creation. Even so, Gemini 4 Argon has no confirmed release date in Brazil, while SynthID depends on an app to work and is not available in the country. Google is swapping the traditional AI approach for specialized models and integrating AI chips into satellites, redefining efficiency in complex tasks and resilience in extreme environments. Gemini 4 Argon sets new records in software development and cybersecurity, demonstrating advances in processing power and accuracy in technical tasks. Project Suncatcher will test the resilience of AI chips in extreme space conditions, paving the way for applications in hostile environments and over the long term.