Google has announced that Gemini 4 Carbon, its most powerful AI model, is now officially available to developers, enterprises, researchers, and consumers. The model was previously offered on a limited basis to cybersecurity professionals through a special program, and users across various sectors can now access or try it through the corresponding channels. Gemini 4 Carbon delivers outstanding performance across multiple mainstream benchmark tests, striking a balance between frontier-level accuracy and deployment costs, with the API early-access discount running through November 30. The model is designed for highly challenging tasks such as complex code debugging and open-ended scientific research. It features multimodal knowledge processing, multi-agent collaboration, cybersecurity protection, and scientific research reasoning capabilities, and can support highly complex scenarios including 3D content generation, complex code refactoring, end-to-end vulnerability handling, and cryptographic protocol code verification. The existing free plan for researchers has also been expanded to include the model.