Google Debuts New AI Models, Genie 4 and Gemini 4 Carbon: What You Need to Know

Google has unveiled two new AI models, Gemini 4 Carbon and Genie 4, just two weeks after its previous release. The company's latest large language model, Gemini 4 Carbon, is its most advanced yet, surpassing its previous model, Gemini 4 Argon, as well as competitors such as OpenAI's GPT-6 Astra, in benchmarking tests that measure a model's overall capabilities.

However, Google is positioning Gemini 4 Carbon primarily for cybersecurity professionals, noting its ability to detect and address vulnerabilities, a skill that was demonstrated by uncovering significant PHP vulnerabilities recently. The company's other new model, Genie 4, is a world model capable of simulating both real and imagined physical worlds, with the addition of audio capabilities that allow developers to incorporate background noise and dialogue into their AI-generated environments.

As of now, neither model is accessible to the general public, with only selected cybersecurity experts having access to Gemini 4 Carbon. Google declined to comment on the anticipated timeline for public access to these models, but based on past releases, it may be within one to two weeks. Google is currently navigating the US government's review process regarding these new AI models.

The company emphasizes its ongoing engagement with regulators. While Google remains a major player in the AI industry, it has lagged behind competitors in releasing new models in an industry that sees frequent updates. Public concern surrounding AI, including fears of job losses, misinformation, and the creation of deepfakes, has also intensified in recent years.