---
title: TomoroAI/tomoro-colqwen3-embed-4b
description: TomoroAI/tomoro-colqwen3-embed-4b is a state-of-the-art ColPali-style multimodal embedding model. It maps text queries, visual documents (im. Qwen3-VL, 4.4B parameters.
canonical_url: https://superlinked.com/models/tomoroai-tomoro-colqwen3-embed-4b
last_updated: 2026-08-24
---

# TomoroAI/tomoro-colqwen3-embed-4b

TomoroAI/tomoro-colqwen3-embed-4b is a state-of-the-art ColPali-style multimodal embedding model. It maps text queries, visual documents (images, PDFs) or short videos into aligned multi-vector embeddings.

Source: [TomoroAI/tomoro-colqwen3-embed-4b on HuggingFace](https://huggingface.co/TomoroAI/tomoro-colqwen3-embed-4b)
Base model: [Qwen/Qwen3-VL-4B-Instruct](https://huggingface.co/Qwen/Qwen3-VL-4B-Instruct)

## Overview

| Field | Value |
|-------|-------|
| Architecture | Qwen3-VL |
| Parameters | 4.4B |
| Tasks | Encode |
| Outputs | Multi-Vec |
| Dimensions | Multi-Vec: 320 |
| Max sequence length | 8,192 tokens |
| License | apache-2.0 |
| Inputs | text, image |
| Languages | multilingual |

## Benchmarks

### Vidore3ComputerScienceRetrieval

Domain: technology · Task: retrieval · Language: en

Visual document retrieval on computer science papers and slides

**Variant: default_lang-eng**

**Quality:** map at 10: 0.6491 · mrr at 10: 0.8847 · ndcg at 10: 0.7690

**Variant: muvera_lang-eng_limit-256**

**Quality:** ndcg at 10: 0.8558 · map at 10: 0.7873 · mrr at 10: 0.9134

[Reference](https://arxiv.org/abs/2601.08620)

### Vidore3FinanceEnRetrieval

Domain: finance · Task: retrieval · Language: en

Visual document retrieval on financial reports

**Variant: default_lang-eng**

**Quality:** map at 10: 0.5816 · mrr at 10: 0.8172 · ndcg at 10: 0.6970

**Variant: muvera_lang-eng_limit-256**

**Quality:** ndcg at 10: 0.7161 · map at 10: 0.6000 · mrr at 10: 0.7717

[Reference](https://arxiv.org/abs/2601.08620)

### Vidore3HrRetrieval

Domain: general · Task: retrieval · Language: en

Visual document retrieval on HR-related documents

**Quality:** ndcg at 10: 0.5851 · map at 10: 0.4758 · mrr at 10: 0.6364

[Reference](https://arxiv.org/abs/2601.08620)

### Vidore3PharmaceuticalsRetrieval

Domain: medical · Task: retrieval · Language: en

Visual document retrieval on pharmaceutical documents

**Variant: default_lang-eng**

**Quality:** map at 10: 0.5754 · mrr at 10: 0.7978 · ndcg at 10: 0.6852

**Variant: muvera_lang-eng_limit-256**

**Quality:** ndcg at 10: 0.7916 · map at 10: 0.6890 · mrr at 10: 0.8737

[Reference](https://arxiv.org/abs/2601.08620)