Gemini 4 Carbon: Our next frontier model

Google DeepMind has announced Gemini 4 Carbon, a frontier model the company describes as its strongest yet, with a focus on cybersecurity and long-running engineering work. In a post dated October 28, 2026, Koray Kavukcuoglu, Google DeepMind SVP and Google's Chief AI Architect, said Carbon arrives two weeks after Gemini 4 Argon. Google says it ranks first on version 5.0 of the Artificial Analysis Intelligence Index with a score of 59 and reached #4 on WebDev Arena after an 89-point elo gain.

On cybersecurity, the company reports 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation and 71.0% on CWE-bench for patching, all measured without guardrails. It also claims more than double Argon's success rate at generating secure code on internal tests, and says Chrome's team found Carbon over 10 times better than another lab's large model at producing quality patches.

Google also reports safety gains: successful cheating on Terminal-Bench 4.0 fell to 0.0%, high-severity misalignment alerts in internal coding environments dropped almost 80%, and attack success on Gray Swan's prompt injection benchmark was 0.0%, versus 0.7% for Argon.

Access is limited for now. A subset of cyber defenders gets the model first, with more trusted partners added through the Fairwind Program, while Google participates in the U.S. government's voluntary pre-release testing. Broader availability will follow further guardrail work, with no date given. Introductory pricing matches Argon at $2 per million input tokens and $10 per million output tokens. The restricted rollout underlines how seriously Google treats a model this capable at exploitation.