Google releases Gemini 4 Carbon, its strongest model, to the public after two weeks with security defenders

Google has made Gemini 4 Carbon available to developers, enterprises, scientists and consumers. Koray Kavukcuoglu, SVP at Google DeepMind and Google's Chief AI Architect, announced the release. Google calls Carbon its strongest model and says it performs at the frontier in cybersecurity and in long-horizon agent workflows across knowledge work, software engineering and scientific reasoning. Two weeks earlier, Google gave Carbon to a group of cybersecurity defenders through its Fairwind Program. Carbon adds to the recently launched Gemini 4 Argon rather than replacing it. Google says Argon is still the better choice for most software engineering and knowledge work, while Carbon is meant for the hardest tasks, such as complex multi-step coding and debugging and open-ended scientific research. The Decoder notes that Carbon uses more tokens than Argon. Google also previewed a "/swarm" feature in Antigravity that splits a single prompt across a coordinated team of agents working in real time. The Decoder adds that the release came late: Anthropic's new Fable 5.2 already beats Carbon on the Artificial Analysis benchmark, even at a lower reasoning setting. For customers, the choice between Google's top models now comes down to how hard a task is and how much they are willing to spend on it.