Skip to main content
QwenCloud home page
Docs
Multimodal Embeddings

DashScope multimodal embedding

Multimodal embedding API

POST
https://ap-southeast-1.maas.qwencloudapi.com/api/v1/services/embeddings/multimodal-embedding/multimodal-embedding
# Singapore (ap-southeast-1). For another region, replace ap-southeast-1 in the URL.
# API keys are region-specific. See https://docs.qwencloud.com/developer-guides/getting-started/regions
curl --location --request POST \
  'https://ap-southeast-1.maas.qwencloudapi.com/api/v1/services/embeddings/multimodal-embedding/multimodal-embedding' \
--header "Authorization: Bearer $DASHSCOPE_API_KEY" \
--header 'Content-Type: application/json' \
--data '{
    "model": "tongyi-embedding-vision-plus",
    "input": {
        "contents": [
            {"text": "Multimodal embedding model"},
            {"image": "https://example.com/image.jpg"},
            {"video": "https://example.com/video.mp4"}
        ]
    }
}'
{
  "output": {
    "embeddings": [
      {
        "index": 0,
        "embedding": [
          0
        ],
        "type": "text"
      }
    ]
  },
  "usage": {
    "input_tokens": 0,
    "input_tokens_details": {
      "image_tokens": 0,
      "text_tokens": 0
    },
    "output_tokens": 0,
    "total_tokens": 0,
    "image_tokens": 0
  },
  "request_id": "1fff9502-a6c5-9472-9ee1-73930fdd04c5"
}
Convert text, images, and video into numerical vectors in a unified semantic space for cross-modal retrieval, similarity search, and content classification.
Before you begin: get an API key, set it as an environment variable, and install the DashScope SDK if you use the SDK.

Endpoint

This API's models are available only in Singapore (ap-southeast-1) and China North 2 (Beijing). The following uses the Singapore endpoint; for a Beijing-only model, replace ap-southeast-1 with cn-beijing and use an API key from Beijing. See Regions and endpoints.
  • HTTP: POST https://ap-southeast-1.maas.qwencloudapi.com/api/v1/services/embeddings/multimodal-embedding/multimodal-embedding
  • SDK base_url: https://ap-southeast-1.maas.qwencloudapi.com/api/v1

Model overview

tongyi-embedding-vision-plus and tongyi-embedding-vision-flash are available only in Singapore (ap-southeast-1). qwen3-vl-embedding and multimodal-embedding-v1 are available only in China North 2 (Beijing): call https://cn-beijing.maas.qwencloudapi.com with an API key created in that region.
  • Singapore
  • China North 2 (Beijing)
ModelModalitiesDimensionsText limitImage limitVideo limit
tongyi-embedding-vision-plusText, Image, Video, Multi-images1152 (default), 1024, 512, 256, 128, 641,024 tokens3 MB per image, max 8 images10 MB
tongyi-embedding-vision-flashText, Image, Video, Multi-images768 (default), 512, 256, 128, 641,024 tokens3 MB per image, max 8 images10 MB
Only qwen3-vl-embedding supports fused vectors (enable_fusion). qwen3-vl-embedding, tongyi-embedding-vision-plus, and tongyi-embedding-vision-flash accept the dimension parameter. The other models have fixed dimensions and return one independent vector per input item.
The res_level (0, 1, 2, or 3; default 1) and max_video_frames (default 8, max 64) parameters are supported only by tongyi-embedding-vision-plus-2026-03-06 and tongyi-embedding-vision-flash-2026-03-06. res_level sets the input resolution level (127, 402, 578, or 1026 tokens per image). max_video_frames sets the maximum number of sampled video frames.

Notes

  • Image input: Public URL or Base64 data URI (data:image/{format};base64,{data}). tongyi-embedding-vision-* and multimodal-embedding-v1 accept JPG, PNG, and BMP; qwen3-vl-embedding also accepts WEBP, TIFF, ICO, DIB, ICNS, and SGI.
  • Multi-images: Key multi_images, supported only by tongyi-embedding-vision-plus and tongyi-embedding-vision-flash. Value is a list of image URLs, max 8 images. Each image must be no larger than 3 MB.
  • Video input: Must be a public URL. tongyi-embedding-vision-* and multimodal-embedding-v1 accept MP4, MPEG, AVI, MOV, MPG, WEBM, FLV, and MKV; qwen3-vl-embedding accepts MP4, AVI, and MOV. Use the fps parameter in parameters to control frame sampling rate (range [0, 1], default 1.0). Values outside this range do not cause an error.
  • Items per request: tongyi-embedding-vision-* has no item count limit (total tokens must stay within the text limit). qwen3-vl-embedding accepts up to 20 items, of which at most 5 are images. multimodal-embedding-v1 accepts up to 20 items: at most 1 image, 1 video, and 20 text entries.

Authorizations

​
string
header
required

DashScope API Key. Create one in the QwenCloud console. Alternatively, you can pass the API Key via the X-DashScope-ApiKey request header.

Body

application/json
​
enum<string>
required

Model name for multimodal embedding. tongyi-embedding-vision-plus and tongyi-embedding-vision-flash are available only in Singapore (ap-southeast-1). qwen3-vl-embedding and multimodal-embedding-v1 are available only in China North 2 (Beijing) (cn-beijing). Use the endpoint and an API key from the model's region.

Available options:tongyi-embedding-vision-plus,tongyi-embedding-vision-flash,qwen3-vl-embedding,multimodal-embedding-v1
Example:tongyi-embedding-vision-plus
​
object
required

Input data containing the content items.

​
object

Parameters for multimodal embedding.

Response

200-application/json
​
object
​
object

Token usage statistics. Fields vary by model: tongyi-embedding-vision-* models return input_tokens (combined text and image token count), input_tokens_details, output_tokens, and total_tokens; other models may return different fields — see individual field descriptions.

​
string

Unique request identifier.

Example:1fff9502-a6c5-9472-9ee1-73930fdd04c5