The result as text for a model: text items verbatim, embedded text resources inline, other media described. Falls back to the structured content's JSON.