Holo-3.1-4B

Name: Holo-3.1-4B
Author: Hcompany

by Hcompany

4B vision-language model for GUI agents and computer-use perception

1.3Kdl/month

55likes

4Bparams

HuggingFace Run it on your own data, free

Identifiers

Model ID

Hcompany/Holo-3.1-4B

Feature URI

mixpeek://image_extractor@v1/hcompany_holo_31_4b_v1

Overview

Holo-3.1-4B is a compact vision-language model tagged for action, agent, computer use, and GUI agents. It is relevant to multimodal search because many agent traces are not documents. They are screenshots, browser states, UI elements, and before-after visual states from tool calls.

On Mixpeek, Holo can turn screenshots and UI recordings into searchable agent memory. That lets an agent retrieve prior visual states, inspect similar failures, and compare what the screen looked like before deciding whether to retry, stop, or ask for help.

Architecture

Qwen-family image-text-to-text model with Hugging Face metadata for action, agent, computer use, GUI agents, and conversational visual reasoning.

Mixpeek SDK Integration

// No extractor parameter takes a Hugging Face model id (checked against
// GET /v1/discovery/extractors, which returns 13), so Holo-3.1-4B runs
// on your side and the output is upserted through POST
// /v1/namespaces/{namespace_id}/documents/upsert. On Enterprise the other
// path is to upload the weights instead: POST /v1/namespaces/{id}/models
// accepts the huggingface format and a custom plugin loads them.
const res = await fetch(
  "https://api.mixpeek.com/v1/namespaces/ns_your_namespace/documents/upsert",
  {
    method: "POST",
    headers: {
      Authorization: "Bearer API_KEY",
      "Content-Type": "application/json",
    },
    body: JSON.stringify({
      collection_id: "col_your_collection",
      documents: [
        {
          document_id: "asset-00412",
          // The model produces text, so it lands in payload. Give the
          // collection a text vector index and embed that text to make it
          // searchable rather than only filterable.
          payload: { extracted_text: modelOutput, source_key: "archive/2026/asset-00412" },
          vectors: { "image-embedding": embeddingOfModelOutput },
        },
      ],
    }),
  },
);

// Managed alternative, if this exact model is not the requirement:
// image_extractor@v1 runs google/siglip-base-patch16-224
// (768-d) over a bucket, with no inference of your own.