Texte vers Embeddings

Converts text to a numeric embedding vector. Useful for vector storage or similarity comparison.

Parameters

Parameter	Type	Required	Variable	Description
`inputArray`	dynamic value	Yes	Yes	Array of items containing the text to embed.
`textField`	text	Yes	No	Name of the field containing text in each array item.
`model.provider`	choice (`openai`, `cohere`, `ollama`, `local`)	No	No	Embedding model provider. (Default: `"openai"`)
`model.model`	text	No	No	Embedding model identifier to use. (Default: `"text-embedding-3-small"`)
`batchSize`	number	No	No	Number of items processed per batch for embedding. (Default: `50`, min 1, max 2048)
`dimensions`	number	No	No	Generated vector size (depends on model, e.g. 1536 for OpenAI). (min 64, max 4096)
`outputField`	text	No	No	Field name where the vector will be stored in each item. (Default: `"embedding"`)
`outputVariable`	text	No	No	Output variable name containing items with their vectors.

Parameters marked Variable = Yes accept the {{blockName.field}} syntax.

Output variable : embeddingsResult

{
  "success": false,
  "results": "...",
  "totalItems": 0,
  "embeddedItems": 0,
  "dimension": 0,
  "provider": "...",
  "durationMs": 0
}

Generate an embedding for a text.

Input :

{"text": "intelligence artificielle"}

Output :

{"embedding": [0.012, -0.034, ...], "dimensions": 1536}

Tip

The returned vector has the embedding model dimension (1536 by default). Combine with a vector storage block to index your data.