On 28 October 2026, Google announced Gemini 4 Carbon, its strongest model to date featuring state-of-the-art cybersecurity and long horizon capabilities. This new release follows the launch of Gemini 4 Argon just two weeks prior and aims to deliver major gains across coding and agentic tasks.
Key takeaways
- Gemini 4 Carbon ranks number one on version 5.0 of the Artificial Analysis Intelligence Index with a score of 59.
- Google reports strong performance in cybersecurity benchmarks including SEC-bench Pro, ExploitGym, and CWE-bench.
- The model launches at the same introductory price as Argon, charging $2 per million input tokens and $10 per million output tokens.
- Access is currently limited to a subset of cyber defenders and trusted partners through the Fairwind Program before a broader rollout.
What is Gemini 4 Carbon?
Google reports that Gemini 4 Carbon is its newest frontier model built with advanced security and long horizon capabilities. The system ranks number one on the updated v5.0 of the Artificial Analysis Intelligence Index at a score of 59. It brings major improvements to coding and agentic tasks, showing particular strength in defending digital networks.
On internal benchmarks, the model drives a strong uplift in generating secure code. When tested in live coding environments by Google teams, it showed reduced misalignment and fabrication rates compared to earlier versions.
How does google frontier model perform in cybersecurity?
According to Google, the new release lands a firm lead on several key vulnerability benchmarks. It achieved 88.2 percent on SEC-bench Pro for finding vulnerabilities, 54.8 percent on ExploitGym for exploitation, and 71.0 percent on CWE-bench for patching code errors.
The tech giant is taking a deliberate approach to releasing the system. Testing includes active engagement with the United States government voluntary process for pre-release model evaluation.
What happens next?
Starting this week, Google is making the model available to a subset of cyber defenders. Access will phase out gradually to additional trusted partners through the Fairwind Program. Feedback will help teams iterate on guardrails before developers, businesses, and consumers can use the technology across product surfaces. Pricing is set at $2 per million input tokens and $10 per million output tokens, with cached input tokens discounted by 95 percent.
Frequently asked questions
Gemini 4 Carbon is Google's strongest AI frontier model featuring state-of-the-art cybersecurity and long horizon capabilities, announced on 28 October 2026.
Google reports that Carbon ranks number one on the recently updated version 5.0 of the Artificial Analysis Intelligence Index at a score of 59.
Google states that the model lands a firm lead in SEC-bench Pro at 88.2 percent for finding vulnerabilities, ExploitGym at 54.8 percent for exploitation, and CWE-bench at 71.0 percent for patching.
The model launches at the same introductory price as Argon, costing $2 per million input tokens and $10 per million output tokens, with cached input tokens priced at 95 percent off.
Google is making the model available to a subset of cyber defenders and phasing access to additional trusted partners through the Fairwind Program before a broader public rollout.
Based on reporting by Google. Read the original article
