Google is back with another new frontier AI model, just two weeks after its last release. The new large language model is dubbed Gemini 4 Carbon, the company announced on Wednesday. And if you’re a fan of world models, there’s good news there, too: Google’s latest, Genie 4, also debuted this week.
Gemini 4 Carbon is the most advanced the company has released. It surpassed its own Gemini 4 Argon model and competitors like OpenAI’s GPT-6 Astra on benchmarking tests designed to assess a model’s overall capabilities. But Google is pitching the model as one meant for cybersecurity experts. Google says Gemini 4 Carbon is good at detecting and patching vulnerabilities — a skill we’ve already seen can uncover major problems, as with last week’s PHP vulnerabilities affecting a majority of the world’s websites.
Google’s other new AI model this week is Genie 4. This is a world model, which can simulate real and imagined physical worlds. Whereas chatbots predict the next word in a sentence, world models are grounded in the physics of a digital universe. They’re useful for things like robotics and video game design. Genie 4 will introduce a new capability: audio. Developers will be able to add background noise and dialogue to their AI worlds. You may have heard of two AI bigwigs, Fei-Fei Li and Yann LeCun, building world models in their respective start-ups.
More from CNET
Neither of these new models, Gemini 4 Carbon nor Genie 4, is available to the general public yet. Right now, only select cybersecurity experts have access to Gemini 4 Carbon. Google declined to answer a question from CNET about when we can expect access to open up, but if we go by previous releases, it may be as soon as one to two weeks from now. Google says it is actively engaging with the US government’s review process. You can join the waitlist for Genie 4 now.
Google ended 2025 as a major player in the AI industry, but it hasn’t kept up with its peers in the so-called “AI race.” Gemini 4 Argon was the company’s first new frontier model in 11 months — a long time in an industry that sees new models seemingly every week.
But public appetite for AI has changed a lot since 2025. Fears around AI have risen, as has public backlash against AI infrastructure projects such as data centers. While fears about AI contributing to human extinction are often overblown or actually give tech companies more power, other fears about AI leading to mass job losses, misinformation and deepfakes are alive and well.