Intelligent multimedia delivery uses semantic information about content to enhance the delivery process. This paper proposes a model for intelligent multimedia delivery that advances the state of the art by incorporating a concept of semantic distortion into the delivery optimization process. Furthermore, the model combines format-independence with rate-distortion optimization to provide a flexible framework for intelligent delivery of multimedia in existing and future formats. The paper provides a review of the existing work covering components of the model: scalable coding formats, rate-distortion optimization, format-independent adaptation, and semantic adaptation. It then details the model, and identifies open problems and opportunities for further research.