`ai_embed`

Generate text embeddings via OpenAI, Ollama, or OAuth-authenticated providers.

Parameters

providerstringdefault "openai": Embedding provider: "openai", "ollama", "oauth"
modelstringdefault per provider: Model name (see defaults below)
input_keystringrequired: Context key holding text (string or array of strings)
output_keystringdefault "embed": Prefix for output context keys
timeoutnumberdefault 120: HTTP request timeout in seconds
api_keystring*: API key (OpenAI)
base_urlstring*: Base URL for OpenAI or OAuth endpoint
ollama_hoststring: Ollama server URL
token_urlstring*: OAuth token endpoint
client_idstring*: OAuth client ID
client_secretstring*: OAuth client secret
scopestring: OAuth scope

*Required for the respective provider; falls back to environment variables.

Default Models

Provider	Default Model
`openai`	`text-embedding-3-small`
`ollama`	`nomic-embed-text`
`oauth`	`openai-text-embedding-3-small`

Environment Variable Fallbacks

Config Key	Environment Variable	Provider
`api_key`	`OPENAI_API_KEY`	openai
`base_url`	`OPENAI_BASE_URL`	openai
`ollama_host`	`OLLAMA_HOST`	ollama
`token_url`	`OAUTH_TOKEN_URL`	oauth
`client_id`	`OAUTH_CLIENT_ID`	oauth
`client_secret`	`OAUTH_CLIENT_SECRET`	oauth
`scope`	`OAUTH_SCOPE`	oauth
`base_url`	`OAUTH_BASE_URL`	oauth

Context Output

Key	Type	Description
`{output_key}_embeddings`	array	Array of embedding vectors (each is array of f64)
`{output_key}_count`	number	Number of embeddings returned
`{output_key}_dimension`	number	Dimension of each embedding vector
`{output_key}_model`	string	Model name used
`{output_key}_success`	boolean	`true` on success

Input

The input_key context value can be:

A string — embedded as a single text, returns one embedding
An array of strings — batch embedded, returns one embedding per text

Examples

OpenAI embeddings

flow:step("embed", nodes.ai_embed({
    provider = "openai",
    model = "text-embedding-3-small",
    input_key = "text",
    output_key = "result"
}))

Ollama (local)

flow:step("embed", nodes.ai_embed({
    provider = "ollama",
    model = "nomic-embed-text",
    input_key = "text",
    output_key = "result"
}))

OAuth-authenticated endpoint

flow:step("embed", nodes.ai_embed({
    provider = "oauth",
    model = "openai-text-embedding-3-small",
    input_key = "text",
    output_key = "result"
}))

Batch embedding

flow:step("prepare", function(ctx)
    return {
        sentences = {
            "The quick brown fox",
            "jumped over the lazy dog",
            "and ran into the forest"
        }
    }
end)

flow:step("embed", nodes.ai_embed({
    provider = "openai",
    input_key = "sentences",
    output_key = "vectors"
})):depends_on("prepare")