Google announced Gemini 4 Carbon as its strongest model. It leads on the Artificial Analysis Intelligence Index v5.0 at 59. It tops SEC-bench Pro at 88.2% for finding vulnerabilities. It scores 54.8% on ExploitGym for exploitation. It reaches 71.0% on CWE-bench for patching. It improves secure code generation by over 2x internally. Chrome reports over 10x better patch quality than another lab’s model. It cuts high-severity misalignment alerts by almost 80%. It resists every evaluated prompt injection attack. It launches at $2 per million input tokens and $10 per million output tokens. Cached input tokens are 95% off. Access starts with a subset of cyber defenders via Fairwind. It will expand to developers, businesses and consumers later.
What changed
Gemini 4 Argon was the latest frontier model.
What it unlocks
Running long-horizon engineering tasks and cybersecurity defense workflows end to end.
- 59 on Artificial Analysis Intelligence Index v5.0
- 88.2% on SEC-bench Pro
- 54.8% on ExploitGym
- 71.0% on CWE-bench
What you need to act on it
- access via Fairwind Program
- U.S. government pre-release testing
Sources