Google launches Gemini 4 Argon to simulate, train and control robots
Google released Gemini 4 Argon to developers, offering a single multimodal model that plans, simulates and controls robots. The system builds virtual practice worlds from photos, videos and sketches, then runs thousands of training episodes to improve real-robot success rates and speed. Argon is available through Google’s paid Gemini API with a one million token combined input and output limit.
Where this came from. 7 outlets reported this. We read them, then wrote the sixty words above.