Google's Gemini 4 Argon spots healthcare security flaw previous AI models missed
The model found a serious weakness in hospital software that earlier AI models had missed.
Google announced Gemini 4 Argon, a new AI model for complex thinking tasks. The company is rolling it out first to cybersecurity professionals through its Fairwind Program.
Argon excels at software engineering, legal research, financial analysis, and cybersecurity defense. In an early test, Argon found a serious weakness.
It exposed personal data across hospital software worldwide. Earlier models had missed this flaw.
The model can process up to one million tokens. Earlier versions could not do this.
This allows it to reason through longer and harder problems. Google prices the model at $2 per million input tokens and $10 per million output tokens.
- Gemini 4 Argon
- Model name
- 1 million tokens, up from 64K
- Token capacity
- $2 per million
- Input token price
- $10 per million
- Output token price
Why it mattersAI systems that reliably find security flaws could help companies protect critical systems like hospitals. They could defend against cyberattacks. This would make hospitals safer.
AustraliaNo direct Australian impact yet, but if Argon helps secure hospitals and critical infrastructure globally, Australian systems could benefit.
✓ Claims checked against the source. checked 2 h ago
Open this story in InSnip →
