Google launches Gemini 4 Carbon with top cybersecurity scores
The new frontier model tops AI benchmarks and is rolling out first to cyber defenders.
28 Oct 2026 0 reads Compiled with AI
On October 28, 2026, Google announced Gemini 4 Carbon, its newest frontier model. It follows Argon, which launched two weeks earlier. Google said its researchers and models worked closely together to advance capabilities faster while tightening safeguards.
Performance and Security Gains
Gemini 4 Carbon ranks first on the v5.0 Artificial Analysis Intelligence Index with a score of 59. It also leads external cybersecurity tests: 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation and 71.0% on CWE-bench for patching. In web development, it gained 89 Elo points to rank fourth on WebDev Arena.
In internal tests, the model shows a strong uplift in generating secure code. In live coding environments with Google's own teams, it also shows reduced misalignment and fabrication rates.
On transparency, the model cuts unsupported claims by about 2x compared with Argon. It states methodological caveats directly in its output and surfaces discrepancies to users before acting. It also reduces overreach by 26% and under-service by 29%. On the Indirect Prompt Injection benchmark, it recorded a 0.0% attack success rate, resisting every evaluated attempt.
The model will first be available to a select group of cyber defenders this week. Google said it will then phase in access to more trusted partners through the Fairwind Program, before bringing Gemini 4 Carbon to developers, businesses and consumers as soon as possible. Introductory pricing matches Argon: $2 per million input tokens and $10 per million output tokens, with cached input tokens 95% off the input price. Afterward, the price rises to $4 per million input tokens and $20 per million output tokens.