Google has introduced Gemini 4 Carbon, a new artificial intelligence model designed to strengthen cybersecurity and improve software development. Developed by Google DeepMind, the model focuses on identifying software vulnerabilities, evaluating security risks and generating code fixes.

Unlike general-purpose AI models, Gemini 4 Carbon places particular emphasis on helping cybersecurity professionals detect and address weaknesses in complex software systems.

The model is designed to support different stages of software security, including:

Google reported strong results across several cybersecurity benchmarks, demonstrating the model's capabilities in identifying and fixing software vulnerabilities.

Google also reported that Gemini 4 Carbon achieved more than twice the success rate of Gemini 4 Argon in generating secure code during internal evaluations.

Google is initially providing access to a selected group of cybersecurity professionals through its Fairwind Program. The company plans to gradually expand access to additional trusted partners before making the model available more widely to developers, businesses and consumers.

Its introductory API pricing is set at $2 per million input tokens and $10 per million output tokens, matching Gemini 4 Argon's introductory pricing.

The launch highlights how AI is becoming an important part of modern cybersecurity. By combining vulnerability detection, security testing and code remediation, Gemini 4 Carbon aims to help development teams identify problems earlier and reduce the effort required to fix them.

However, AI-generated security findings and patches still require independent testing and human review before being used in production systems.

Sources