Google Unveils Gemini 4 Carbon, Its Most Powerful AI Model Yet, Surpassing Claude Opus 5.5 and GPT-6 Astra on the Artificial Analysis Intelligence Index

Google has introduced Gemini 4 Carbon, its latest frontier AI model, two weeks after the release of Gemini 4 Argon. The company described Carbon as its strongest model, with state-of-the-art performance in cybersecurity and long-horizon tasks. It ranks first on the updated Artificial Analysis Intelligence Index v5.0 with a score of 59. Artificial Analysis, an independent benchmarking organization, placed Carbon ahead of Claude Opus 5.5 (58) and GPT-6 Astra (57). Google stated that Carbon leads on multiple evaluations within the index, including complex reasoning, knowledge, and scientific tasks. The model also shows gains in coding and agentic capabilities compared with Argon. Cybersecurity Performance Carbon posts strong results on cybersecurity benchmarks. Google reported 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching. Internal tests indicated that Carbon generates secure code at more than twice the success rate of Argon. The Chrome team reported that Carbon performed more than 10 times better than another lab’s frontier model when generating high-quality patches. Google said the model reduces misalignment and fabrication rates in internal coding environments. It is being made available first to a limited group of cybersecurity defenders through Google’s Fairwind Program while the company continues testing safeguards. Pricing and Availability Gemini 4 Carbon launches at the same introductory pricing as Argon: $2 per million input tokens and $10 per million output tokens. Cached input tokens are priced at 95% off the input rate. After the introductory period, prices are expected to rise to $4 per million input tokens and $20 per million output tokens. Broader availability for developers, businesses, and consumers is expected later. Google has not provided a firm public timeline beyond the initial Fairwind rollout. The model follows the same staged-release approach used for Argon. Competitive Context The launch places Google at the top of a major independent intelligence ranking for the first time in the current cycle. Margins over Anthropic’s Claude Opus 5.5 and OpenAI’s GPT-6 Astra are narrow. On some coding-agent evaluations, Claude Opus 5.5 remains ahead. Carbon leads on most of the 17 benchmarks Google disclosed in its comparison table. The release continues a rapid cadence of frontier model updates from Google DeepMind. Carbon is positioned for long-running engineering tasks and defensive cybersecurity applications. What to Watch Further independent benchmark results as Carbon becomes more widely available. Performance of the model in the Fairwind Program and any reported security findings. When general access opens for developers and enterprise customers. Pricing after the introductory period ends. Comparisons with future releases from Anthropic, OpenAI, and other labs. Adoption in coding, agentic workflows, and cybersecurity defense. Google has unveiled Gemini 4 Carbon as its most powerful AI model, ranking first on the Artificial Analysis Intelligence Index v5.0 with a score of 59 and leading several cybersecurity benchmarks. Initial access is limited to the Fairwind Program, with introductory pricing matching Gemini 4 Argon.