Google introduces Gemini 4 Carbon with leading AI benchmark performance

The new AI model excels in detecting software vulnerabilities, with plans for broader access

Google has introduced its latest artificial intelligence (AI) model, Gemini 4 Carbon, which has achieved the top position on the Artificial Analysis Intelligence Index with a score of 59. Announced on October 29, 2026, the model demonstrates enhanced capabilities in identifying software vulnerabilities, testing their exploitability, and developing solutions.

Initially, Gemini 4 Carbon will be available to a select group of cybersecurity defenders. Google plans to expand access through its Fairwind Program to trusted partners, with eventual availability to developers, businesses, and consumers after further feedback and safeguard improvements.

Google CEO Sundar Pichai highlighted the model’s release on X, stating, “I did say we’d be releasing our models periodically… introducing Gemini 4 Carbon, our next frontier model.”

The model follows the recent launch of Gemini 4 Argon and is designed to manage complex engineering tasks over extended periods. Google DeepMind reports that Carbon can address technical challenges for days, including identifying and fixing bugs across extensive codebases and low-level systems.

Gemini 4 Carbon achieved scores of 88.2% on SEC-bench Pro for vulnerability detection, 54.8% on ExploitGym for exploit testing, and 71% on CWE-bench for developing fixes. Google claims that Carbon’s success rate in generating secure code is more than double that of Argon, with significant improvements reported by its Chrome team.

In terms of reliability and safety, Carbon has reduced unsupported claims by half compared to Argon and decreased high-severity misalignment alerts by nearly 80% in internal testing. It also recorded a zero percent success rate on the Gray Swan indirect prompt-injection benchmark.

The model is priced at USD 2 per million input tokens and USD 10 per million output tokens, with a 95% discount on cached input tokens. Google intends to gradually expand access while continuing to test and enhance the model’s safeguards before a wider release.

Source: https://gulfnews.com/technology/google-unveils-gemini-4-carbon-with-top-ai-benchmark-score-expands-focus-on-cybersecurity-1.500727714