InnoDeep's DNA is innovation and creativity and we also care about responsible and ethical AI. We have contributed to the AI ethics recommendations, conference: UNESCO. 41st General Assembly in 2021 by the amendment No. 126 concerning AI and Health. In this article, we present our new framework, Green MLOps by InnoDeep ™, for eco-responsible artificial intelligence. Indeed, it is an implementation of a multi-scale MLOps solution: Code, infrastructure and data. As we are concerned about the robustness of our artificial intelligence we have set up a non linear cost function with three parameters: Energy balance, AI efficiency and Data/Algorithms. The next challenge is to optimize this function to ensure at the same time a robust and eco-responsible AI. This architceture is being experimented by InnoDeep. In terms of infrastructure, the question arises as to how to optimize under-exploited resources while ensuring high availability for customers.