Google announces the introduction of two new AI models

To further expand the application of artificial intelligence in the tech world, Google has announced the introduction of two new models: 'Gemini 4 Carbon' and 'Genie 4'. One of these models is designed to assist cybersecurity experts in identifying vulnerabilities within computer systems, while the other possesses the capability to create digital worlds based on both real and imaginary environments.

Gemini 4 Carbon: Built for Cybersecurity

According to Google, Gemini 4 Carbon is the company's most advanced AI model to date. The company claims that in benchmark tests, it has outperformed several other models, including Gemini 4 Argon and OpenAI’s GPT-6 Astra.

This model has been specifically developed to support professionals working in the field of cybersecurity. It will help improve the process of identifying and remediating security vulnerabilities within computer systems.

Genie 4: Capable of Creating Digital Worlds

Google’s second model, Genie 4, is a 'world model' primarily designed to construct digital environments that can simulate real or imagined worlds.

While traditional chatbots focus on predicting words in response to user queries, 'world models' aim to understand and replicate environments, the objects within them, and the interactions that take place between those objects. Potential applications for this technology include fields such as robotics, virtual environments, and video game development—areas where there is a need to make digital environments more dynamic and true-to-life.

Audio Features Included in Genie 4

Genie 4 also incorporates audio capabilities, allowing developers to integrate background sounds and character dialogue into the digital worlds they create. This feature helps make these environments feel more realistic and lifelike.

Currently Unavailable to General Users

Google has clarified that neither of the new models is currently available to the general public. Access to 'Genie 4 Carbon' is presently being granted only to select cybersecurity experts; no further details have been released regarding when these models might become widely available to general users and developers.

It is worth noting that Google announced these new models just two weeks after its previous AI-related breakthrough.