Google's Gemini 4 Carbon Tops Artificial Analysis Intelligence Index

Google DeepMind's latest model, Gemini 4 Carbon, has secured the top position on the Artificial Analysis Intelligence Index v5 with a score of 59, surpassing its closest competitor, Anthropic’s Claude Opus 5.5, by one point. This model was unveiled on October 28, 2026, just weeks after the release of its predecessor, Gemini 4 Argon, which scored 50.

The Intelligence Index is a composite score derived from nine evaluations, and Gemini 4 Carbon excels in four of these, including HLE-Diamond and AutomationBench-AA. Notably, it demonstrates impressive capabilities in cybersecurity, achieving an 88.2% score on SEC-bench Pro for vulnerability discovery and 71.0% on CWE-bench for code patching.

Gemini 4 Carbon's performance indicates a significant advancement in AI, particularly for enterprise security teams facing challenges with unpatched code and limited resources. The model is currently accessible to select cybersecurity defenders through Google’s Fairwind Program, with a public API release expected in the future. Initial pricing is set at $4 per million input tokens and $20 per million output tokens, with a 50% discount during the launch period.

As the AI landscape continues to evolve, the performance of Gemini 4 Carbon, especially in coding and vulnerability detection, will be closely monitored, particularly in comparison to its competitors.