A marketing application may start with text-only content generation, then face a new request: use product images alongside written content. A multimodal model can process or generate across modalities such as text, images, audio, and video. That broader capability is useful when the product experience genuinely needs more than text—not merely because the model offers it. section 5.3.5
Written content plus product images
Clients could supply copy and product images, then receive descriptions, social media posts, or suggested image edits. For an initial product focused mainly on high-quality, verifiable text, a multimodal model may be more than is needed; visual features could become a later differentiator. section 5.3.5
Before choosing a multimodal model, check:
- Which modalities the problem and available data require. Text, visual, and auditory data call for different techniques, and model inputs must ultimately be numeric. section 3.1.1
- Whether combining modalities improves the user experience enough to justify the added infrastructure. Handling multiple data streams increases memory, processing, and storage demands compared with a text-only model. section 5.3.5
