Google Unveils Gemini 4 Carbon AI Model Focused on Cybersecurity and Coding

Alphabet's Google has unveiled Gemini 4 Carbon, its latest frontier artificial intelligence model. The launch intensifies competition with industry peers like OpenAI and Anthropic as technology companies race to develop more capable coding and cybersecurity systems.

The announcement follows the introduction of Gemini 4 Argon two weeks prior, marking a continued expansion of Google's Gemini family of AI models tailored for developers and enterprise customers.

According to the company, the Carbon model delivers state-of-the-art performance across several cybersecurity tasks, including vulnerability discovery, system exploitation, and software patching. Internal benchmarks indicate that Gemini 4 Carbon generates secure code at more than twice the success rate of the previously released Argon model. Additionally, the new model is designed to handle complex, long-running engineering and coding tasks.

Google is initially making Gemini 4 Carbon available to a subset of cybersecurity defenders through its Fairwind Program. This controlled rollout allows the company to continue testing essential safeguards before a broader release.

In terms of pricing, Gemini 4 Carbon will be available at $2 per million input tokens and $10 per million output tokens, matching the introductory pricing structure of the Gemini 4 Argon model.

"Google's launch of Gemini 4 Carbon highlights the rapid commercialization of specialized AI models for enterprise security and software development. By targeting critical operational areas like vulnerability discovery and secure code generation, tech giants are moving beyond general-purpose tools to solve specific industry pain points. For businesses, the integration of such advanced models into cybersecurity protocols could significantly alter how threat detection and software patching are managed at scale." — Dr. Shishir Gupta, Founder & CEO, StartupLanes