Implementation:Cohere ai Cohere python EmbedContent Union

Knowledge Sources	Cohere Python SDK
Domains	SDK, Embeddings
Last Updated	2026-02-15 14:00 GMT

Overview

EmbedContent is a discriminated union type that represents the different content types (text or image URL) that can be provided as input to the Cohere Embed API.

Description

The EmbedContent type is a discriminated union (tagged union) composed of two variant classes:

• ImageUrlEmbedContent: Represents an image input identified by the type discriminator value "image_url". Contains an optional image_url field of type EmbedImageUrl (which wraps a base64 URL string).
• TextEmbedContent: Represents a text input identified by the type discriminator value "text". Contains an optional text field of type str.

The union uses the type field as the discriminant, annotated via UnionMetadata(discriminant="type"). This allows the SDK to automatically deserialize the correct variant based on the type value in the API response or request payload.

Both variant classes extend UncheckedBaseModel and support Pydantic v1 and v2 compatibility.

Usage

Use EmbedContent when building multimodal embedding requests that may include both text and image inputs. This union type allows mixing text and image content in a single embed API call.

Code Reference

Source Location

• Repository: Cohere Python SDK
• File: src/cohere/types/embed_content.py

Signature

class ImageUrlEmbedContent(UncheckedBaseModel):
    type: typing.Literal["image_url"] = "image_url"
    image_url: typing.Optional[EmbedImageUrl] = None

class TextEmbedContent(UncheckedBaseModel):
    type: typing.Literal["text"] = "text"
    text: typing.Optional[str] = None

EmbedContent = typing_extensions.Annotated[
    typing.Union[ImageUrlEmbedContent, TextEmbedContent],
    UnionMetadata(discriminant="type"),
]

Import

from cohere.types import EmbedContent
from cohere.types.embed_content import ImageUrlEmbedContent, TextEmbedContent

I/O Contract

Union Variants

Variant	Discriminator Value	Description
ImageUrlEmbedContent	"image_url"	Image content for embedding via a base64 URL
TextEmbedContent	"text"	Text content for embedding

ImageUrlEmbedContent Fields

Field	Type	Required	Default	Description
type	Literal["image_url"]	Yes	"image_url"	Discriminator field identifying this as an image URL content
image_url	Optional[EmbedImageUrl]	No	None	The image URL object containing a base64-encoded image URL

TextEmbedContent Fields

Field	Type	Required	Default	Description
type	Literal["text"]	Yes	"text"	Discriminator field identifying this as text content
text	Optional[str]	No	None	The text string to embed

Usage Examples

Creating Text Embed Content

from cohere.types.embed_content import TextEmbedContent

text_content = TextEmbedContent(
    type="text",
    text="What is machine learning?",
)

print(text_content.type)  # "text"
print(text_content.text)  # "What is machine learning?"

Creating Image URL Embed Content

from cohere.types.embed_content import ImageUrlEmbedContent
from cohere.types import EmbedImageUrl

image_content = ImageUrlEmbedContent(
    type="image_url",
    image_url=EmbedImageUrl(url="data:image/png;base64,iVBORw0KGgo..."),
)

print(image_content.type)          # "image_url"
print(image_content.image_url.url) # "data:image/png;base64,iVBORw0KGgo..."

Multimodal Embed Request

import cohere
from cohere.types.embed_content import TextEmbedContent, ImageUrlEmbedContent
from cohere.types import EmbedImageUrl

co = cohere.Client()

inputs = [
    TextEmbedContent(text="A photo of a sunset over the ocean"),
    ImageUrlEmbedContent(
        image_url=EmbedImageUrl(url="data:image/jpeg;base64,/9j/4AAQ...")
    ),
]

response = co.embed(
    model="embed-english-v3.0",
    input_type="search_document",
    embedding_types=["float"],
    inputs=inputs,
)

print(f"Generated {len(response.embeddings.float_)} embeddings")

Related Pages

• Environment:Cohere_ai_Cohere_python_Python_SDK_Runtime