The Pulse
Google Launches Gemini 4 Carbon With Cybersecurity Focus
Google DeepMind announced Gemini 4 Carbon on October 28, 2026, positioning it as its strongest model for cybersecurity and long-running engineering tasks. The company says Carbon will first go to a limited group of cyber defenders, with bro
AI.info Team · Written with AI · fact-checked ·
Google DeepMind announced Gemini 4 Carbon on October 28, calling it the company’s strongest model for cybersecurity and long-running engineering work. The model will first reach a limited group of cyber defenders, with broader access planned for developers, businesses and consumers later. “Today, we’re announcing the next frontier model: Gemini 4 Carbon, our strongest model with state-of-the-art cybersecurity and long horizon capabilities,” wrote Koray Kavukcuoglu, SVP of Google DeepMind and Google’s chief AI architect.
“Today, we’re announcing the next frontier model: Gemini 4 Carbon, our strongest model with state-of-the-art cybersecurity and long horizon capabilities.”
Koray Kavukcuoglu, SVP, Google DeepMind and Chief AI Architect, Google
Carbon takes the top spot on Artificial Analysis’ index
Google says Carbon scores 59 on version 5.0 of the Artificial Analysis Intelligence Index. Artificial Analysis independently lists Gemini 4 Carbon (High) at 59, ahead of Claude Opus 5.5 (Max, Default Fallback) at 58. The index combines nine evaluations covering areas including mathematics, science, coding and reasoning; Artificial Analysis says Carbon has no publicly available API.
Google also reports gains in coding and agentic tasks, and a score of 1,769 on WebDev Arena, where it says Carbon ranks fourth after an 89-point increase. The company frames the model as a follow-on to Gemini 4 Argon, which launched two weeks earlier. That places Carbon’s initial case less in a new consumer feature than in harder software tasks and security work.
Three security scores, with a deployment caveat
Google reports Carbon scored 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for testing exploitability, and 71.0% on CWE-bench for patching. The post says these external benchmark results were evaluated without guardrails. The scores therefore describe model performance under those evaluation conditions, not how Carbon behaves with Google’s full safety protections active.
The company also cites internal tests: it says Carbon more than doubles Argon’s success rate at generating secure code, while Google’s Chrome team found it performed more than ten times better than an unnamed frontier model at producing high-quality patches. Google does not identify that comparison model or provide the test details in the announcement. Those claims are company-reported results, distinct from the independently maintained Artificial Analysis index.
Fairwind partners get access first
Google plans to begin access this week with a subset of cyber defenders, then add trusted partners through its Fairwind Program. The company says it is taking part in the U.S. government’s voluntary process for pre-release model testing and will continue refining guardrails before widening access. It has not announced a date for general availability.
Google lists the same introductory API price as Argon: $2 per million input tokens and $10 per million output tokens, with cached input tokens discounted by 95%. After the introductory period, the listed rates rise to $4 per million input tokens and $20 per million output tokens. For now, the announced access remains limited to cyber defenders and selected partners.