MiniCPM-V-4.6

Name: MiniCPM-V-4.6
Author: openbmb

by openbmb

1B-parameter edge VLM that matches 2B-class quality on vision tasks

222Kdl/month

1B total (0.8B language + 0.4B vision)params

HuggingFace Run it on your own data, free

Identifiers

Model ID

openbmb/MiniCPM-V-4.6

Feature URI

mixpeek://image_extractor@v1/openbmb_minicpm_v46_v1

Overview

MiniCPM-V-4.6 is a 1B-parameter multimodal language model from OpenBMB designed for deployment on mobile and edge devices. Built on Qwen3.5-0.8B with a SigLIP2-400M vision encoder, it achieves performance comparable to models twice its size on vision-language benchmarks. It supports image understanding, video comprehension (up to 128 frames), OCR, and tool calling: all within a footprint that runs on smartphones.

Architecture

Frozen-tower vision-language model combining a SigLIP2-400M image encoder with a Qwen3.5-0.8B language decoder. Uses mixed 4x/16x visual token compression to balance detail and efficiency. Supports arbitrary image resolutions via dynamic tiling. Video input processes up to 128 frames with temporal position encoding.

Mixpeek SDK Integration

// No extractor parameter takes a Hugging Face model id (checked against
// GET /v1/discovery/extractors, which returns 13), so MiniCPM-V-4.6 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: { "multimodal-embedding": embeddingOfModelOutput },
        },
      ],
    }),
  },
);

// Managed alternative, if this exact model is not the requirement:
// universal_extractor@v1 runs google/gemini-embedding-2
// (3072-d) over a bucket, with no inference of your own.