Google Unveils Gemini 4 Argon With Advanced Safeguards Against Misuse

Hindustan Times · technology

Google has introduced its new frontier AI model, Gemini 4 Argon, designed for complex workflows in areas like software engineering, finance, and cybersecurity. The model boasts an impressive one million token limit. Before its release, Google implemented robust safeguards, including refusing harmful requests, enhancing resilience against prompt injections, and deploying mitigations to monitor and control the AI's actions. These measures aim to prevent misuse and ensure the AI operates within intended parameters. The model is priced at $2 per million input tokens and $10 per million output tokens.