Google's Gemini 4 Carbon promises 'significant leap' in AI coding and cybersecurity

Google has unveiled its latest frontier model, Gemini 4 Carbon, which claims to outperform competitors in several vital benchmarks, particularly in the realms of cybersecurity and coding. This new model builds upon the recent launch of Gemini 4 Argon, signifying the rapid advancement being made in AI technology.

According to a company blog post, Gemini 4 Carbon has achieved the highest AA Index v5.0 score of 59, just slightly ahead of Claude Opus 5.5 and GPT-6 Astra. Google also asserts that its model showcases dominance across a range of categories, including agentic coding, computer use, knowledge work, science, and mathematics.

The cybersecurity category is where Google particularly highlights the strength of Gemini 4 Carbon. In this area, the model excels in identifying and addressing vulnerabilities at their root cause, offering a significant advantage over other models. Google claims that the model is twice as likely to generate secure code compared to its predecessor, Gemini 4 Argon.

Initially, Gemini 4 Carbon will be made available to a select group of cybersecurity defense companies for pre-release model testing. Google intends to broaden access to trusted partners via its Fairwind Program before releasing the model to a wider audience.