Skip to navigation
Request capabilitiesMultimodal

Images

Send images to vision-capable models by URL or base64

Send images with your prompt to have a model describe, compare, or extract from them. Use image_url for a URL or data: URI, or image for raw base64; Gateway converts either to the served vendor’s format.

curl https://api-gateway.merge.dev/v1/responses \
  -H "Authorization: Bearer $MERGE_GATEWAY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-5.1",
    "input": [{
      "type": "message",
      "role": "user",
      "content": [
        {"type": "text", "text": "Compare these two product shots."},
        {"type": "image_url", "url": "https://example.com/before.jpg"},
        {"type": "image_url", "url": "https://example.com/after.jpg"}
      ]
    }]
  }'

With the OpenAI SDK at /v1/openai, send the standard {"type": "image_url", "image_url": {"url": "..."}} part, or input_image on Responses. AI SDK file parts take data as bytes, base64, a data: URI, or a URL.

AI SDK file parts need merge-gateway-ai-sdk-provider 0.3.0 or later; earlier versions drop them silently.

Reference

FieldNotes
image_url.urlAn https:// URL or a data: URI
image.source_typebase64 (default) or url
image.media_typeRequired, even for a URL, such as image/png
image.dataBase64 string, or the URL when source_type is url

Inline images are capped at 20 MB decoded per request, all images combined; URLs don’t count. Over the cap returns 413 payload_too_large, and a route without image input returns 400 unsupported_params.

Next steps

Documents

Send PDFs to document-capable models

Image generation

Generate images from a prompt