Google announces Gemini 4 models and advancements in AI simulation

Google has released a report detailing significant advancements in its artificial intelligence ecosystem. The company introduced Gemini 4 Argon, a model designed for complex reasoning in fields such as programming, financial analysis, and legal work. This is followed by Gemini 4 Carbon, which focuses on long-term task execution and cybersecurity. Carbon is currently undergoing limited testing with the US government and select cybersecurity professionals.

Other major developments include Genie 4, an AI model capable of generating real-time, interactive virtual worlds with physical simulations. Google also expanded its SynthID Detector to help identify AI-generated images, video, and audio. In scientific research, the company reported successful AI processing in space using a TPU on a low-earth orbit satellite to run the Gemma 4 model. Additionally, Google announced progress in quantum computing, using its Willow processor to reduce errors in generating specific quantum states by approximately 40 times.