Skip to content

Models

Multimodal concepts

Working with images, audio, and documents.

Some models accept inputs beyond text, such as images or audio, and some produce non-text outputs. Each model page lists supported input and output modalities.

Inputs

  • Images are passed as URLs or base64 data within message content.
  • Audio inputs are used by transcription and speech models.
  • Large documents may need to be split to fit the context window.