Google Expands Gemini 4 Carbon Access to Developers and Consumers

Google has expanded access to its Gemini 4 Carbon AI model, making it available to developers, enterprises, scientists, and consumers, two weeks after initially launching it for cybersecurity professionals. The model reportedly outperforms Google's previous Argon model and competitor models from OpenAI and Anthropic in benchmark tests, with particular strengths in knowledge work, agentic coding, science, mathematics, and cybersecurity. It also features improved spatial reasoning and physics simulations for 3D game engines and creative coding. Google emphasizes safety, noting the model achieved a 0% prompt-injection attack success rate on the Gray Swan IPI benchmark. Gemini 4 Carbon is accessible via AI Studio and Android Studio for developers, while AI Ultra subscribers can use it through the Gemini App.