AI & agents
Google unveils Gemini 4 Carbon, its next frontier model
- Published
- Source
- Google blog
Summary
Google announced Gemini 4 Carbon on October 28, 2026. According to the company, it targets state-of-the-art cybersecurity and long-horizon capabilities, with large gains in coding and agentic tasks. Figures such as 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym, 71.0% on CWE-bench for patching, and the top score (59) on the Artificial Analysis Intelligence Index v5.0 are all Google's own reported results. It builds on Gemini 4 Argon, released two weeks earlier.
Why it matters for our work
It widens the model options for long-running agent work and security tasks such as vulnerability analysis. Since the numbers are vendor-reported, teams should verify them in their own workflows before adopting.
Translated from the Korean original. Summaries may be translated and edited. Commentary reflects our perspective; forecasts remain the source’s views.