Given that the provider of a general-purpose AI model may not know all possible downstream uses of the model, whether within or outside the scope of this Regulation, the provider should provide sufficient information to permit the downstream provider of an AI system to understand the general-purpose AI model and properly integrate it into their AI system. For the same reason, the provider of a general-purpose AI model should provide sufficient information about its training and testing process, including, as appropriate, summary results of evaluations, to allow the downstream AI system provider and relevant competent authority to understand the key characteristics, capabilities and limitations of the model. This information should support the downstream provider in their own compliance activities under this Regulation, in particular the AI system provider's activities in relation to high-risk AI systems.