Gemini 4 Argon has only been out for 15 days and Google already dropped a more powerful model.
- TB 4.0: 57.1% to 65.2%
- TB Science: 52.9% to 64.8%
- ExploitGym: 42.0% to 54.8%
- GraphWalks: 84.2% to 96.2%
- SWE-sweep: 4.3% to 9.6%
顯示更多
Gemini 4 Carbon is our next frontier model.
It’s state-of-the-art across cybersecurity and long-horizon capabilities.