Embed v4 (embed-multimodal-v4) Logo

Embed v4 (embed-multimodal-v4)

Provider: Cohere

Cohere’s most advanced embedding model to date. Produces unified vector representations from mixed input types—text, image, PDFs, slides—with “Matryoshka” multi-resolution outputs. :contentReference[oaicite:8]{index=8}

Key Information:

  • Identifier: embed-multimodal-v4
  • Fine-tunable: Yes
  • Standard Model: No