The EmbedClient.V2Client class provides methods to create embeddings synchronously for multimodal content, returning embeddings immediately in the response.

Methods

Create sync embeddings

Description: This method synchronously creates embeddings for multimodal content and returns the results immediately in the response.

Use this method to embed a query for retrieving matching content. With Marengo 3.5, audio and video can be up to 30 seconds. With Marengo 3.0, they can be up to 10 minutes. For longer content, use the embed.v_2.tasks.create method instead.

The content this method accepts depends on the model. With Marengo 3.5, this method accepts only the multi_input input type; provide text, images, audio, or video as media sources. With Marengo 3.0, use the individual input types. For the formats, resolutions, file sizes, and duration limits each model accepts, see the input requirements for Marengo 3.5 or Marengo 3.0.

Note

This method is rate-limited. With Marengo 3.5, the platform counts input tokens for each type of content. A request can exceed a limit before you see an error. For details, see Input token limits for embedding.

Function signature and example

def create(
    self,
    *,
    input_type: CreateEmbeddingsRequestInputType,
    model_name: CreateEmbeddingsRequestModelName,
    auto_truncate: typing.Optional[bool] = OMIT,
    embedding_uncertainty: typing.Optional[bool] = OMIT,
    embedding_dimension: typing.Optional[int] = OMIT,
    text: typing.Optional[TextInputRequest] = OMIT,
    image: typing.Optional[ImageInputRequest] = OMIT,
    text_image: typing.Optional[TextImageInputRequest] = OMIT,
    audio: typing.Optional[AudioInputRequest] = OMIT,
    video: typing.Optional[VideoInputRequest] = OMIT,
    multi_input: typing.Optional[MultiInputRequest] = OMIT,
    request_options: typing.Optional[RequestOptions] = None,
) -> EmbeddingSuccessResponse

Parameters

Name	Type	Required	Description
`input_type`	`CreateEmbeddingsRequestInputType`	Yes	The type of content for the embeddings. Values: - `multi_input`: Text and up to 10 media sources, combined into a single embedding. To reference a specific media source from your text, use a placeholder in the format `<@name>`, where `name` matches the `name` field of a media source. Marengo 3.5 accepts images, video, audio, and documents as media sources. Marengo 3.0 accepts images. - `audio`: An audio file. Requires Marengo 3.0. - `video`: A video file. Requires Marengo 3.0. - `image`: An image file. Requires Marengo 3.0. - `text`: Text input. Requires Marengo 3.0. - `text_image`: Text and an image. Requires Marengo 3.0.
`model_name`	`CreateEmbeddingsRequestModelName`	Yes	The embedding model to use. Values: - `marengo3.5`: For details about this version, see the Marengo 3.5 page. - `marengo3.0`: For details about this version, see the Marengo 3.0 page.
`text`	`TextInputRequest`	No	Text input configuration. Required when `input_type` is `text`. See TextInputRequest for details.
`image`	`ImageInputRequest`	No	Image input configuration. Required when `input_type` is `image`. See ImageInputRequest for details.
`text_image`	`TextImageInputRequest`	No	Combined text and image input configuration. Required when `input_type` is `text_image`. See TextImageInputRequest for details.
`audio`	`AudioInputRequest`	No	Audio input configuration. Required when `input_type` is `audio`. See AudioInputRequest for details.
`video`	`VideoInputRequest`	No	Video input configuration. Required when `input_type` is `video`. See VideoInputRequest for details.
`multi_input`	`MultiInputRequest`	No	Text and media source configuration. Required when `input_type` is `multi_input`. See MultiInputRequest for details.
`auto_truncate`	`Optional[bool]`	No	Controls the behavior of the platform when the text in your request exceeds 2,000 tokens. Requires Marengo 3.5. Values: - `False`: The platform returns a `400` error. - `True`: Truncate your text to fit the limit, and set the `usage.truncated` field to `True` in the response.
`embedding_uncertainty`	`Optional[bool]`	No	Set this parameter to `True` to include a per-dimension uncertainty vector in the `data[].embedding_uncertainty` field of the response. The vector has the same length as the `embedding` array. A higher value indicates lower confidence in that dimension. Requires Marengo 3.5. Requirements: - Set this parameter to `True` only for a text-only or media-only request. If you combine text with media sources, the platform returns a `400` error. - The platform returns a `400` error if your request includes a document, whether PDF, plain text, or Markdown.
`embedding_dimension`	`Optional[int]`	No	The number of dimensions for each embedding in the response, including the `data[].embedding_uncertainty` vector. Marengo 3.5 produces Matryoshka embeddings: a shorter embedding consists of the first values of the full-length embedding. A 256-dimension embedding, for example, is the first 256 values of a 512-dimension embedding of the same content. Shorter embeddings reduce index size and speed up similarity search; longer embeddings produce higher retrieval quality. Requirements: - Requires Marengo 3.5. Setting this parameter when `model_name` is `marengo3.0` returns a `400` error. - Applies to the entire request: you cannot set it for a single input type or embedding. - Use the same value across an index. Default: 512
`request_options`	`RequestOptions`	No	Request-specific configuration.

TextInputRequest

The TextInputRequest class specifies the configuration for processing text content. Required when input_type is text.

Name	Type	Required	Description
`input_text`	`str`	Yes	The text for which you wish to create an embedding. The maximum length is 500 tokens.

ImageInputRequest

The ImageInputRequest class specifies the configuration for processing image content. Required when input_type is image. Requires Marengo 3.0. The decoded file can be up to 32 MB.

Name	Type	Required	Description
`media_source`	`MediaSource`	Yes	Specifies the source of the image file. See MediaSource for details.

TextImageInputRequest

The TextImageInputRequest class specifies the configuration for processing combined text and image content. Required when input_type is text_image. Requires Marengo 3.0. The decoded file can be up to 32 MB.

Name	Type	Required	Description
`media_source`	`MediaSource`	Yes	Specifies the source of the image file. See MediaSource for details.
`input_text`	`str`	Yes	The text for which you wish to create an embedding. The maximum length is 500 tokens.

AudioInputRequest

The AudioInputRequest class specifies the configuration for processing audio content. Required when input_type is audio. Requires Marengo 3.0. The decoded file can be up to 36 MB.

Name	Type	Required	Description
`media_source`	`MediaSource`	Yes	Specifies the source of the audio file. See MediaSource for details.
`start_sec`	`float`	No	The start time in seconds for processing the audio file. Use this parameter to process a portion of the audio file starting from a specific time. Default: 0 (start from the beginning).
`end_sec`	`float`	No	The end time in seconds for processing the audio file. Use this parameter to process a portion of the audio file ending at a specific time. The end time must be greater than the start time. Default: End of the audio file
`segmentation`	`AudioSegmentation`	No	Specifies how the platform divides the audio into segments. See AudioSegmentation for details.
`embedding_option`	`List[str]`	No	The types of embeddings you wish to generate. Values: - `audio`: Generates embeddings based on audio content (sounds, music, effects) - `transcription`: Generates embeddings based on transcribed speech You can specify multiple values to generate different types of embeddings for the same audio. Default: `["audio", "transcription"]`
`embedding_scope`	`List[str]`	No	The scope for which you wish to generate embeddings. Values: - `clip`: Generates one embedding for each segment - `asset`: Generates one embedding for the entire audio file You can specify multiple scopes to generate embeddings at different levels. Default: `["clip", "asset"]`
`embedding_type`	`List[str]`	No	Specifies how to structure the embedding. Include this parameter only when the `embedding_option` parameter contains at least two values. Values: - `separate_embedding`: Returns separate embeddings for each modality specified in the `embedding_option` parameter. - `fused_embedding`: Returns a single combined embedding that integrates all modalities into one vector. Specify both values to receive separate and fused embeddings in the same response. Default: `separate_embedding`.

VideoInputRequest

The VideoInputRequest class specifies the configuration for processing video content. Required when input_type is video. Requires Marengo 3.0. The decoded file can be up to 36 MB.

Name	Type	Required	Description
`media_source`	`MediaSource`	Yes	Specifies the source of the video file. See MediaSource for details.
`start_sec`	`float`	No	The start time in seconds for processing the video file. Use this parameter to process a portion of the video file starting from a specific time. Default: 0 (start from the beginning)
`end_sec`	`float`	No	The end time in seconds for processing the video file. Use this parameter to process a portion of the video file ending at a specific time. The end time must be greater than the start time. Default: End of the video file
`segmentation`	`VideoSegmentation`	No	Specifies how the platform divides the video into segments. See VideoSegmentation for details.
`embedding_option`	`List[str]`	No	The types of embeddings to generate for the video. Values: - `visual`: Generates embeddings based on visual content (scenes, objects, actions) - `audio`: Generates embeddings based on audio content (sounds, music, effects) - `transcription`: Generates embeddings based on transcribed speech You can specify multiple values to generate different types of embeddings for the same video. Default: `["visual", "audio", "transcription"]`
`embedding_scope`	`List[str]`	No	The scope for which you wish to generate embeddings. Values: - `clip`: Generates one embedding for each segment - `asset`: Generates one embedding for the entire video file. Use this scope for videos up to 10-30 seconds to maintain optimal performance. You can specify multiple scopes to generate embeddings at different levels. Default: `["clip", "asset"]`
`embedding_type`	`List[str]`	No	Specifies how to structure the embedding. Include this parameter only when the `embedding_option` parameter contains at least two values. Values: - `separate_embedding`: Returns separate embeddings for each modality specified in the `embedding_option` parameter. - `fused_embedding`: Returns a single combined embedding that integrates all modalities into one vector. Specify both values to receive separate and fused embeddings in the same response. Default: `separate_embedding`.

MultiInputRequest

The MultiInputRequest class specifies the configuration for processing text and media sources. Required when input_type is multi_input.

Marengo 3.5 accepts images, video, audio, and documents as media sources. Marengo 3.0 accepts images.

Include text in the input_text field when you combine media sources of different types. For example, if you combine an image and a video without text, the platform returns a 400 error. Media sources of the same type do not require text.

Document sources

The platform embeds a plain text or Markdown document as text. Combining content into a single embedding requires at least one image, video, or audio source; if the content is all text, the platform returns a 400 error. A plain text or Markdown document combined with the input_text field, and two plain text documents, are both all-text content. Send a single text document as your only source, use input_text on its own, or add an image, video, or audio source.

A PDF document must be your only source. You cannot combine it with any other media source, including another document, or with the input_text field; the platform returns a 400 error if you do. To combine document content with an image, video, or audio source, send a plain text or Markdown document instead of a PDF file.

Name	Type	Required	Description
`input_text`	`str`	No	Text to include in the embedding. With Marengo 3.5, the text cannot exceed 2,000 tokens. Use the `auto_truncate` parameter to control the behavior of the platform when your text exceeds it. Usage options: - Provide text without media sources to create a text-only embedding. - Combine text with media sources to add context. Example: “A person cooking.” - Use media source references to describe relationships between specific media sources. The format is `<@name>`, where `name` matches the `name` field of a media source. Example: “A person wearing <@outfit> and holding <@accessory>.” - Omit this field to create an embedding from media sources only.
`media_sources`	`List[MultiInputMediaSource]`	No	An array of up to 10 media sources to include in the embedding. Omit it to create a text-only embedding from the `input_text` field. The platform processes media sources in the order they appear in the array. If you use media source references in the `input_text` parameter, each must have a corresponding media source with a matching `name` field. If a reference has no match, the request fails.

MediaSource

The MediaSource class specifies the source of the media file. Provide exactly one of the following:

Name	Type	Required	Description
`base_64_string`	`str`	No	The base64-encoded media data. Encoding grows the payload by about a third, so the string you send is larger than the original file. The maximum size depends on the input type and the model. The description of the field that contains this media source states the limit where it differs; for the formats and sizes each model accepts, see the input requirements for Marengo 3.5 or Marengo 3.0.
`url`	`str`	No	The publicly accessible URL of the media file. Use direct links to raw media files. Video hosting platforms and cloud storage sharing links are not supported.
`asset_id`	`str`	No	The unique identifier of an asset from a direct or multipart upload. The asset status must be `ready`. Use `assets.retrieve` to check the status.

MultiInputMediaSource

A class specifying a media source for multi-input embeddings. You must provide exactly one of the url, base_64_string, or asset_id fields. With Marengo 3.5, each media source can be up to 32 MB, whichever of the three fields you use. Audio and video can be up to 30 seconds. If your content exceeds either limit, the platform returns either a 400 or a 413 error. A PDF file also has a page allowance: 16 pages for each MB of file size. A 0.5 MB file is allowed 16 pages, and a 4 MB file is allowed 64 pages. The platform checks the page count of the file against the allowance before processing the file. If the file exceeds the allowance, the platform returns a 413 error. The error message includes the page count and the allowance. Plain text and Markdown files have no page allowance.

Name	Type	Required	Description
`name`	`str`	No	The unique identifier for this media source. This field is required when `input_text` references this media source.
`media_type`	`MultiInputMediaSourceMediaType`	No	The type of media. Values: - `image`: An image file. Works with both Marengo 3.0 and Marengo 3.5. - `video`: A video file. Requires Marengo 3.5. - `audio`: An audio file. Requires Marengo 3.5. - `document`: A PDF (`.pdf`), plain text (`.txt`), or Markdown (`.md`) file. Requires Marengo 3.5. For the rules on combining a document with other sources, see MultiInputRequest.
`url`	`str`	No	The publicly accessible URL of the media file. Use direct links to raw files. Media hosting platforms and cloud storage sharing links are not supported.
`base_64_string`	`str`	No	The base64-encoded media data.
`asset_id`	`str`	No	The unique identifier of an asset from a direct or multipart upload.

AudioSegmentation

The AudioSegmentation class specifies how the platform divides the audio into segments using fixed-length intervals.

Name	Type	Required	Description
`strategy`	`Literal["fixed"]`	Yes	The segmentation strategy. Value: `fixed`.
`fixed`	`AudioSegmentationFixed`	Yes	Configuration for fixed segmentation. This object is required when the `strategy` field is `fixed`. See AudioSegmentationFixed for details.

AudioSegmentationFixed

The AudioSegmentationFixed class configures fixed-length segmentation for audio.

Name	Type	Required	Description
`duration_sec`	`int`	Yes	The duration in seconds for each segment. The platform divides the audio into segments of this exact length. The final segment may be shorter if the audio duration is not evenly divisible. Min: `2`. Max: `10`. Example: With `duration_sec: 5`, a 12-second audio file produces segments: [0-5s], [5-10s], [10-12s].

VideoSegmentation

The VideoSegmentation type specifies how the platform divides the video into segments. Use one of the following:

Fixed segmentation: Divides the video into equal-length segments:

Name	Type	Required	Description
`strategy`	`Literal["fixed"]`	Yes	The segmentation strategy. Value: `fixed`.
`fixed`	`VideoSegmentationFixedFixed`	Yes	Configuration for fixed segmentation. See VideoSegmentationFixedFixed for details.

Dynamic segmentation: Divides the video into adaptive segments based on scene changes:

Name	Type	Required	Description
`strategy`	`Literal["dynamic"]`	Yes	The segmentation strategy. Value: `dynamic`.
`dynamic`	`VideoSegmentationDynamicDynamic`	Yes	Configuration for dynamic segmentation. See VideoSegmentationDynamicDynamic for details.

VideoSegmentationFixedFixed

The VideoSegmentationFixedFixed class configures fixed-length segmentation for video.

Name	Type	Required	Description
`duration_sec`	`int`	Yes	The duration in seconds for each segment. The platform divides the video into segments of this exact length. The final segment may be shorter if the video duration is not evenly divisible. Min: `2`. Max: `10`. Example: With `duration_sec: 5`, a 12-second video produces segments: [0-5s], [5-10s], [10-12s].

VideoSegmentationDynamicDynamic

The VideoSegmentationDynamicDynamic class configures dynamic segmentation for video based on scene changes.

Name	Type	Required	Description
`min_duration_sec`	`int`	Yes	The minimum duration in seconds for each segment. The platform divides the video into segments that are at least this long. Segments adapt to scene changes and content boundaries and may be longer than the minimum. Min: `2`. Max: `5`. Example: With `min_duration_sec: 3`, segments might be: [0-3.2s], [3.2-7.8s], [7.8-12.1s]

Return value

Returns an EmbeddingSuccessResponse object containing the embedding results.

The EmbeddingSuccessResponse class contains the following properties:

Name	Type	Description
`data`	`List[EmbeddingData]`	Array of embedding results.
`usage`	`Optional[EmbeddingUsage]`	Token counts for the request. Only Marengo 3.5 returns this field. See EmbeddingUsage for details.
`metadata`	`Optional[EmbeddingMediaMetadata]`	Metadata for the media input. Available for the `image`, `text_image`, `audio`, `video`, and `multi_input` input types. See EmbeddingMediaMetadata for details.

The EmbeddingData class contains the following properties:

Name	Type	Description
`embedding`	`List[float]`	The embedding vector for the content.
`embedding_uncertainty`	`Optional[List[float]]`	A per-dimension uncertainty vector with the same length as the `embedding` array. A higher value indicates lower confidence in that dimension. Present when the request sets `embedding_uncertainty: true`. Only Marengo 3.5 returns this field.
`embedding_option`	`Optional[EmbeddingDataEmbeddingOption]`	The type of the embedding. Values: - `visual`: Embedding based on visual content (a video, a page of a PDF file, or an image embedded asynchronously). - `audio`: Embedding based on audio content. - `transcription`: Embedding based on transcribed speech. Returned only for content embedded with Marengo 3.0. - `text`: Embedding based on the text content of a PDF, plain text, or Markdown file embedded asynchronously. - `fused`: Embedding based on a combination of the modalities specified in the request. The platform returns this embedding only for video and audio input, and only when the `embedding_type` parameter includes the `fused_embedding` value. - `null`: For text embeddings and images embedded synchronously.
`embedding_scope`	`Optional[EmbeddingDataEmbeddingScope]`	The scope for which the embedding was generated. Values: - `clip`: Embedding for a segment. For video and audio input, one embedding per detected segment. - `page`: Embedding for one page of a PDF file embedded asynchronously, or for one quadrant of a page when the request sets `document.segmentation.spatial.strategy` to `quadrants`. With that strategy, five entries share the same scope and page numbers, so read the `quadrant` field to tell them apart: the whole-page entry has no `quadrant` value. - `chunk`: Embedding for one chunk of whole sentences of a plain text or Markdown file embedded asynchronously. Read the `chunk_index` field for the position of the chunk in the file. - `asset`: Embedding for the entire file. For video and audio input, use this scope for content up to 10-30 seconds to maintain optimal performance. - `null`: For text embeddings and images embedded synchronously.
`start_sec`	`Optional[float]`	The start time in seconds for this segment. This field is `null` for text and image embeddings.
`end_sec`	`Optional[float]`	The end time in seconds for this segment. This field is `null` for text and image embeddings.
`start_page_number`	`Optional[int]`	The first page this embedding covers, counting from 1. The platform returns this field only for page-level embeddings of a PDF file, and `null` in every other case.
`quadrant`	`Optional[str]`	The quarter of the page this embedding covers. The platform returns this field only when the request sets `document.segmentation.spatial.strategy` to `quadrants`, and only on the four quadrant embeddings of a page. This field is `null` on the whole-page embedding and in every other case.
`chunk_index`	`Optional[int]`	The position of this chunk in the file, counting from 0. The platform returns this field only on `chunk`-scope embeddings of a plain text or Markdown file, and `null` in every other case. Read this field rather than the position of the entry in the `data` array, which provides no ordering guarantee.
`end_page_number`	`Optional[int]`	The last page this embedding covers, counting from 1 and including that page. This field matches the `start_page_number` field when the embedding covers a single page. The platform returns this field only for page-level embeddings of a PDF file, and `null` in every other case.

EmbeddingUsage

The EmbeddingUsage class provides token counts for the request. Only Marengo 3.5 returns this object.

Name	Type	Description
`input_tokens`	`Dict[str, int]`	The number of tokens the request used. Each key names a type of content the request processed, and each value is the token count for that content. The platform reports only the content types your request actually used. The possible keys are `video`, `audio`, `image`, `document`, and `text`. Read the keys the response returns rather than assuming a fixed set.
`truncated`	`bool`	Whether the input was truncated to fit within the token limit.

EmbeddingMediaMetadata

The EmbeddingMediaMetadata type provides metadata for the media input. Available for the image, text_image, audio, video, and multi_input input types. The input_type field selects one variant:

Image: Metadata for image embeddings.

Name	Type	Description
`input_type`	`Literal["image"]`	The type of the input content. Value: `image`.
`input_url`	`Optional[str]`	The publicly accessible URL for the image file.
`input_filename`	`Optional[str]`	The name of the image file.

Text and image: Metadata for text-image embeddings.

Name	Type	Description
`input_type`	`Literal["text_image"]`	The type of the input content. Value: `text_image`.
`input_url`	`Optional[str]`	The publicly accessible URL for the image file.
`input_filename`	`Optional[str]`	The name of the image file.

Audio: Metadata for audio embeddings.

Name	Type	Description
`input_type`	`Literal["audio"]`	The type of the input content. Value: `audio`.
`input_url`	`Optional[str]`	The publicly accessible URL for the audio file.
`input_filename`	`Optional[str]`	The name of the audio file.
`embedding_options`	`List[str]`	The `embedding_option` values used to generate the embedding.
`embedding_scopes`	`List[EmbeddingAudioMetadataEmbeddingScopesItem]`	The `embedding_scope` values used to generate the embedding.
`embedding_dimension`	`Optional[int]`	The number of dimensions for each embedding in this response. Only Marengo 3.5 returns this field.
`duration`	`float`	The duration of the audio in seconds.
`start_offset_sec`	`Optional[float]`	The start offset in seconds.
`end_offset_sec`	`Optional[float]`	The end offset in seconds.

Video: Metadata for video embeddings.

Name	Type	Description
`input_type`	`Literal["video"]`	The type of the input content. Value: `video`.
`input_url`	`Optional[str]`	The publicly accessible URL for the video file.
`input_filename`	`Optional[str]`	The name of the video file.
`clip_length`	`Optional[int]`	Length of each video clip in seconds. Only available for fixed segmentation.
`embedding_scopes`	`List[EmbeddingVideoMetadataEmbeddingScopesItem]`	The `embedding_scope` values used to generate the embedding.
`embedding_dimension`	`Optional[int]`	The number of dimensions for each embedding in this response. Only Marengo 3.5 returns this field.
`embedding_options`	`List[str]`	The `embedding_option` values used to generate the embedding.
`duration`	`float`	The duration of the video in seconds.
`start_offset_sec`	`Optional[float]`	The start offset in seconds.
`end_offset_sec`	`Optional[float]`	The end offset in seconds.

Multi-input: Metadata for multi-input embeddings.

Name	Type	Description
`input_type`	`Literal["multi_input"]`	The type of the input content. Value: `multi_input`.
`embedding_dimension`	`Optional[int]`	The number of dimensions for each embedding in this response. Only Marengo 3.5 returns this field.

API Reference

Create sync embeddings

Embed a query

Methods

Create sync embeddings

Parameters

TextInputRequest

ImageInputRequest

TextImageInputRequest

AudioInputRequest

VideoInputRequest

MultiInputRequest

MediaSource

MultiInputMediaSource

AudioSegmentation

AudioSegmentationFixed

VideoSegmentation

VideoSegmentationFixedFixed

VideoSegmentationDynamicDynamic

Return value

EmbeddingUsage

EmbeddingMediaMetadata

API Reference

Related guide