Skip to main content

Technical Information

Known limitations

image-description

  • Multilingual - Yes
  • Max input image size – 5 MB
  • Image formats: JPEG, PNG, GIF, or WebP
  • Maximum total pixels per image: 8000 px * 8000 px. Images will first be scaled down, preserving aspect ratio, until it’s within the size limits.
  • Recommended pixels per image: Less than 1568 pixels on the larger side.
  • Very small images under 200 pixels on any given edge may degrade performance.

image-metadata-generation

  • Multilingual - Yes
  • Max input image size – 5 MB
  • Image formats: JPEG, PNG, GIF, or WebP
  • Maximum total pixels per image: 8000 px * 8000 px. Images will first be scaled down, preserving aspect ratio, until it’s within the size limits.
  • Recommended pixels per image: Less than 1568 pixels on the larger side.
  • Very small images under 200 pixels on any given edge may degrade performance.

text-classification

  • Max input characters - 800K

text-summarization

  • Max input characters - 800K

image-classification

  • Multilingual - Yes
  • Max input image size – 5 MB
  • Image formats: JPEG, PNG, GIF, or WebP
  • Maximum total pixels per image: 8000 px * 8000 px. Images will first be scaled down, preserving aspect ratio, until it’s within the size limits.
  • Recommended pixels per image: Less than 1568 pixels on the larger side.
  • Very small images under 200 pixels on any given edge may degrade performance.

image-embeddings

  • Languages – English
  • Max input image size – 25 MB
  • Image formats: PNG, JPEG
  • Maximum total pixels per image: 204820483
  • Aspect ratio (w/h): min: 0.25, max: 4
  • Output vector size – 1,024

text-embeddings

  • none

named-entity-recognition-image

  • Multilingual - Yes
  • Max input image size – 5 MB
  • Image formats: JPEG, PNG, GIF, or WebP
  • Maximum total pixels per image: 8000 px * 8000 px. Images will first be scaled down, preserving aspect ratio, until it’s within the size limits.
  • Recommended pixels per image: Less than 1568 pixels on the larger side.
  • Very small images under 200 pixels on any given edge may degrade performance.

named-entity-recognition-text

  • Max input characters - 800K