Google Expands Gemini 4 Carbon Availability

Google has announced the expansion of its latest frontier AI model, Gemini 4 Carbon, to a broader audience including developers, enterprises, scientists, and consumers. This development comes two weeks after the initial announcement of the model for cybersecurity professionals. Gemini 4 Carbon demonstrates significant improvements across various domains, particularly in knowledge work, agentic coding, science, mathematics, and cybersecurity.

The model outperforms its predecessor, Gemini Argon, as well as competitors from OpenAI and Anthropic in benchmark tests. It also exhibits enhanced spatial reasoning and physics simulations for 3D game engines and creative coding. Gemini 4 Carbon has been tested with a 0% success rate in prompt-injection attacks during the Gray Swan IPI benchmark test, showcasing its robustness against potential misuse and cyberattacks.

While the everyday user may not fully benefit from the new model, cybersecurity experts, scientists, and other professionals are more likely to leverage the enhancements. Developers can access Gemini 4 Carbon through AI Studio and Android Studio, while users subscribed to the AI Ultra plan can access it via the Gemini App.