Octen Text Embedding Generator

Supply a list of texts and turn them into numeric vectors with Octen. Choose a model, vector length and query/document role if needed. Your Agent can prepare the vectors for later semantic comparison or retrieval.

Your AI

Use with your AI

For tasks with several steps, continue in the AI you already use.

CodexClaude CodeCursorWindsurfCline

View instructions

Use Toolmu's octen-embedding-octen, catalog ID 9120a895-4979-47e1-9bdb-a226f5df7a50, to generate vectors for the texts I provide. Reuse the existing Toolmu connection; if setup is needed, follow https://toolmu.com/SKILL.md. Read get_tool for this exact Tool, check current availability and price, and honor my existing authorization and budget. Ask for the text list if it is missing. Use my selected model, dimension and input_type; use query for a retrieval question or document for passages intended for an index. Keep model and vector length compatible with the collection they will be compared against. Organize the returned vectors with a reliable mapping to their input texts, retain useful returned metadata and mention any truncation warning. Explain the next step needed for my task. Storing vectors, searching an index and preparing an answer from retrieved passages are separate work; report those steps as completed only if you actually perform them.

Task:
Help my AI find the right passage when someone asks a similar question. Prepare [these messages: “Reset your password in account settings”; “Recover access with your registered email”] and keep each message paired with its matching result.

Paste into your connected AI's conversation.

Connection guide

Up to 120 credits / call

The final charge is based on confirmed usage. Unused reserved credits return to your available balance.

Octen

Octen Text Embedding Generator

You receive

When conversion completes, the result contains numeric vectors representing the submitted texts. Each vector is a sequence of numbers, rather than readable prose. Your Agent can link available vectors back to their texts and use a compatible vector index for similarity comparisons or retrieval as a separate step.

Start with your task

Help my AI find the right passage when someone asks a similar question. Prepare [these messages: “Reset your password in account settings”; “Recover access with your registered email”] and keep each message paired with its matching result.

What to know

Vectors do not search the web or answer a question themselves. Text length limits or truncation can affect what is represented. Use compatible model and dimension settings before comparing vectors.

Inputs & outputs

Inspect the current inputs, availability, and maximum charge before starting.

Input schema

Input schema
Field / Path	Type	Requirement	Meaning
Embedding request`body`	`object`	Required	Request object containing the text list and any chosen model, vector length or retrieval role. Required fields `input`
Texts to embed`body.input`	`array`	Required	A list with at least one text entry. Put each passage or search question in its own string so the returned vector can be associated with that text. Minimum items `1`
One text`body.input[]`	`string`	—	One passage or search question to represent as a vector. Keep it concise and split long documents into coherent passages.
Embedding model`body.model`	`string`	Optional	Choose octen-embedding-0.6b, octen-embedding-4b or octen-embedding-8b. The default is octen-embedding-4b. Use a consistent model for documents and queries that will be compared. Default `octen-embedding-4b` Options `octen-embedding-0.6b, octen-embedding-4b, octen-embedding-8b`
Vector length`body.dimension`	`integer`	Optional	Requested number of numeric components per vector. Supply a positive integer supported by the selected model and your index; omission uses the model's maximum length. Maximum `9007199254740991`
Retrieval role`body.input_type`	`string`	Optional	Use query for a retrieval question and document for text intended for your collection. This sets the embedding role; it does not perform the later search. Options `query, document`

View raw JSON

{
  "type": "object",
  "required": [
    "body"
  ],
  "properties": {
    "body": {
      "type": "object",
      "$schema": "https://json-schema.org/draft/2020-12/schema",
      "required": [
        "input"
      ],
      "~standard": {
        "vendor": "zod",
        "version": 1,
        "jsonSchema": {}
      },
      "properties": {
        "input": {
          "type": "array",
          "items": {
            "type": "string"
          },
          "minItems": 1,
          "description": "Texts to embed (max 32768 tokens per element)."
        },
        "model": {
          "enum": [
            "octen-embedding-0.6b",
            "octen-embedding-4b",
            "octen-embedding-8b"
          ],
          "type": "string",
          "default": "octen-embedding-4b"
        },
        "dimension": {
          "type": "integer",
          "maximum": 9007199254740991,
          "description": "Output vector dimensionality (defaults to the model's max).",
          "exclusiveMinimum": 0
        },
        "input_type": {
          "enum": [
            "query",
            "document"
          ],
          "type": "string",
          "description": "Whether the input is a retrieval query or a document."
        }
      },
      "additionalProperties": false
    }
  },
  "additionalProperties": false
}

Result schema

Output fields have not been documented yet.

Additional notes

Keep the original text alongside each vector. The numeric representation is useful for comparing meaning, but it does not replace the passage, its source or factual checking. If the response supplies an input index, use it to maintain that association.

Prepare passages that each carry one useful idea. Character counts and token counts measure different things, so a short-looking passage can still use many tokens. Split long material before embedding it; if a truncation warning is returned, some of the original text may not be represented.

Use a model and vector length compatible with the collection you plan to search. The dimension is the number of numeric components in a vector. A different model can use a different representation, even when vector lengths happen to match.

The query role is for questions to compare with an indexed collection; the document role is for the material in that collection. Selecting a role influences embedding preparation. It does not search a database, discover webpages or retrieve an answer.

Retain useful returned metadata, such as the model or any warning, when supplied. Database storage, similarity calculations, ranking and answer generation are separate Agent or application steps. A similarity value depends on the comparison method and collection; it is not a universal confidence score.

Octen: text embeddings and retrieval preparation

Octen: Embedding request reference

Questions about this tool.

Inputs, results and details to know before running it.

Get connected

A vector is a sequence of numbers used for comparison or indexing. It does not contain a readable explanation or a factual answer. Your Agent can keep the original passage available and prepare a summary separately if that is your task.

Octen Text Embedding Generator

Octen Text Embedding Generator

You receive

Start with your task

What to know

Inputs & outputs

Input schema

Result schema

Choose the text you want to prepare

Say what you will use the text for

Keep each passage with its result

Questions about this tool.

You receive

Start with your task

What to know

Inputs & outputs

Input schema

Result schema

Choose the text you want to prepare

Say what you will use the text for

Keep each passage with its result

Questions about this tool.

Can I read a vector as a summary of the text?

Can vectors from different models or dimensions be mixed?

What if a long passage is truncated?

When should I choose query or document?