Nomic Embed Multimodal 7B

Embedding modelOpen weightsCard summarized

a dense state-of-the-art multimodal embedding model that excels at visual document retrieval tasks

Source: Nomic Embed Multimodal 7B Hugging Face model card (no technical report; launch blog no longer reachable)

Use it via API

Self-host only: no published API price.

Code example Compare

Where to call it

Self-host (weights)nomic-ai/nomic-embed-multimodal-7b

Supports

Image input

Full feature support

Good for

Retrieving document page images directly, skipping OCR
Research papers with equations, diagrams and tables
Technical docs with code blocks, flowcharts and screenshots

Overview

Nomic · 02Multimodal

Embed MM 7B

a dense state-of-the-art multimodal embedding model that excels at visual document retrieval tasks

Typical price

—

No published price

Dimensions: —
Max input: —
Modes: multimodal embeddings, visual document retrieval (page screenshots)
Released: —

Nomic Embed Multimodal 7B is a 7B-parameter dense (single-vector) embedding model fine-tuned from Qwen2.5-VL 7B Instruct. It embeds text, images and interleaved text and images, and is aimed at retrieving visually rich document pages (screenshots of PDFs, slides, reports) without OCR. Nomic reports 59.7 average nDCG@5 on ViDoRe v2. Weights, training data and code are released under Apache 2.0. model card

nomic-ai/nomic-embed-multimodal-7b

Typical price: —
Languages: English, French, German, Italian, Spanish (listed on model card; strongest in English)
Weights: Open · Apache 2.0
Released: —

multimodal embeddings
visual document retrieval (page screenshots)
interleaved text and image embeddings