Get the app

Google Expands Gemini 4 Carbon Availability

CNET
Read full story· 2 min
Google has expanded access to its Gemini 4 Carbon AI model beyond cybersecurity professionals to developers, enterprises, scientists, and consumers. The model excels in knowledge work, coding, science, and cybersecurity, outperforming competitors and showing strong resistance to prompt-injection attacks.

What happened

Google has broadened access to its Gemini 4 Carbon AI model, which was initially launched for cybersecurity professionals two weeks ago. The model is now available to developers, businesses, researchers, and general consumers. Gemini 4 Carbon is designed to handle tasks in areas like coding, scientific research, and cybersecurity. Google has made the model accessible through its AI Studio, Android Studio, and the Gemini App for users on the AI Ultra plan.

Key details

  • Gemini 4 Carbon was first announced for cybersecurity professionals two weeks ago.
  • The model is now open to developers, enterprises, scientists, and consumers.
  • It outperformed Google's Argon model and models from OpenAI and Anthropic in benchmark tests.
  • The model achieved a 0% success rate for prompt-injection attacks on the Gray Swan IPI benchmark.
  • Gemini 4 Carbon is available via AI Studio, Android Studio, and the Gemini App for AI Ultra subscribers.

Why it matters

Developers, scientists, and businesses now have access to a more advanced AI tool for tasks like coding, scientific analysis, and cybersecurity. This expansion gives these groups improved capabilities and stronger protections against prompt-injection attacks. Everyday users subscribed to the AI Ultra plan can also use the model, but the most significant benefits are expected for professionals working with complex data and security challenges.

#anthropic#gemini-4-carbon#google#openai#scientific-research

More in Computer Vision