Create Dataset | Coactive AI

curl -X POST https://api.coactive.ai/api/v1/datasets \
     -H "Authorization: Bearer <token>" \
     -H "Content-Type: application/json" \
     -d '{
  "name": "march_2025_news_clips",
  "encoder_name": "multimodal-tx-large2"
}'

{
  "created_dt": "2026-05-29T19:00:16.311121Z",
  "created_by": "user123",
  "id": "f47ac10b-58cc-4372-a567-0e02b2c3d479",
  "name": "march_2025_sports_clips",
  "embeddings": [
    {
      "id": "44b5df56-fb37-4132-8902-2940c80d9089",
      "name": "multimodal",
      "modality": "image",
      "modalities": [
        "image",
        "video"
      ]
    }
  ],
  "encoders": [
    {
      "id": "44b5df56-fb37-4132-8902-2940c80d9089",
      "name": "multimodal",
      "modality": "image",
      "modalities": [
        "image",
        "video"
      ]
    }
  ],
  "description": "Clips from March Madness",
  "allowed_asset_types": [
    "image",
    "video"
  ],
  "status": "active",
  "keyframe_sampling_method": "intelligent",
  "store_original": true,
  "store_preview": true,
  "updated_dt": "2026-05-29T19:00:16.311172Z",
  "updated_by": "user123",
  "deleted_dt": "2026-05-29T19:00:16.311236Z",
  "deleted_by": "user123",
  "keyframe_uniform_fps_sampling_rate": 1,
  "metadata_fields_to_index": [
    "title",
    "description"
  ],
  "metadata_fields_to_embed": [
    "title",
    "description"
  ],
  "dataset_config": {
    "embedding_id": {
      "audio_transcript": "a8d6a1f6-800d-499a-bb37-822fc71dfac2",
      "audio_transcript_chunk": "c5e1d8a0-3f4b-4c9b-9d1e-2f6a8b5c7d3e",
      "capped-shot-segment": "09bdcde7-8b89-4229-9494-35a4e52690ca",
      "image": "b25cf36e-e9b3-41d2-a856-47c891622cf9"
    },
    "encoder_name": {
      "audio_transcript": "st-qwen-p6b",
      "audio_transcript_chunk": "st-qwen-p6b",
      "capped-shot-segment": "nova2-multimodal-1024",
      "image": "nova2-multimodal-1024"
    },
    "extra_config": {
      "composite_settings": {
        "capped-shot-segment": {
          "maximum_slice_duration_seconds": 30
        }
      }
    },
    "hybrid_search": {
      "weights": {
        "image": 0.5,
        "text": 0.5
      }
    },
    "new_video_data_model": false,
    "context_studio_enabled": false,
    "metadata_fields_to_embed": [
      "title",
      "description"
    ],
    "metadata_fields_to_index": [
      "title",
      "description"
    ],
    "celebrity_detection_enabled": false,
    "agentic_search_metadata_description": {
      "director": "The film director's name",
      "year": "Release year as integer"
    },
    "asset_identifier_key": "external_id",
    "context_studio_export_json_connection_name": "my_connection"
  }
}

Create a new dataset

Authentication

AuthorizationBearer

Bearer authentication of the form Bearer <token>, where token is your auth token.

Request

This endpoint expects an object.

namestringRequiredformat: "^[a-zA-Z0-9_]+$">=1 character

Name of the dataset, must be unique per organization and must only contain alphanumeric and underscore characters.

encoder_nameenumRequired

The model used to extract visual and audio features from assets.

dataset_idstring or nullOptionalformat: "uuid"

A known dataset ID; for internal use only during migration

add_to_permitbooleanOptionalDefaults to true

Flag to use for adding dataset to permit as a resource or not

descriptionstring or nullOptional

A short summary of what the dataset includes or is intended for.

allowed_asset_typeslist of enumsOptional

Types of assets allowed in this dataset.

keyframe_sampling_methodenumOptional

Sampling method to extract keyframes from videos in dataset.

store_originalbooleanOptionalDefaults to true

Whether to store original files.

store_previewbooleanOptionalDefaults to true

Whether to store preview versions.

metadata_fields_to_indexlist of strings or nullOptional

The metadata fields to index for the dataset.

metadata_fields_to_embedlist of strings or nullOptional

The metadata fields to embed for the dataset.

dataset_configobject or nullOptional

keyframe_uniform_fps_sampling_ratedouble or nullOptional0.00001-5Deprecated

If using uniform sampling method, number of keyframes/sec to sample from videos (allowed values: ≥ .00001 to 5 fps).

Response

Successful Response

created_dtstringformat: "date-time"

The created datetime of the resource in ISO 8601 format.

created_bystring

ID of the user who added the asset to the dataset.

idstringformat: "uuid"

The unique identifier for the dataset.

namestring

Name of the dataset, must be unique per organization and must only contain alphanumeric and underscore characters.

embeddingslist of objects

The encoders of the dataset.

encoderslist of objects

The encoders of the dataset.

descriptionstring or null

A short summary of what the dataset includes or is intended for.

allowed_asset_typeslist of enums

Types of assets allowed in this dataset.

statusenum or null

The status of the dataset.

keyframe_sampling_methodenum

Sampling method to extract keyframes from videos in dataset.

store_originalboolean

Whether the original file is stored.

store_previewboolean

Whether preview files are stored.

updated_dtstring or nullformat: "date-time"

The datetime the resource was last updated in ISO 8601 format.

updated_bystring or null

ID of the user who last updated the asset.

deleted_dtstring or nullformat: "date-time"

Timestamp if the resource was deleted.

deleted_bystring or null

ID of the user who deleted the resource.

keyframe_uniform_fps_sampling_ratedouble or null

FPS rate used for uniform sampling.

metadata_fields_to_indexlist of strings or null

The metadata fields to index for the dataset.

metadata_fields_to_embedlist of strings or null

The metadata fields to embed for the dataset.

dataset_configobject or null

Errors

422

Unprocessable Entity Error