# Perplexity's text-only 9B embedding model beats leading retrievers on visual documents

> Original title: The 9B was trained only on text, yet it outperforms the leading dense retrievers on visual documents.

- Source: Perplexity Developers
- Published: 2026-11-02T21:51:04.000Z
- Bad Influence: https://badinfluence.news/items/hpkbmjixqtcyaalatr7lsb3ih
- Original: https://x.com/perplexitydevs/status/2117373956017803306

## Summary

Perplexity says its 9B embedding model, trained only on text, outperforms the leading dense retrievers on visual documents. Using two 1024-dimensional embeddings per page, it beats Nemotron-3.5-Embed-8B on ViDoRe-V3 with a 4x smaller index.

## Post

The 9B was trained only on text, yet it outperforms the leading dense retrievers on visual documents.

With two 1024d embeddings per page, it beats Nemotron-3.5-Embed-8B on ViDoRe-V3 with a 4x smaller index.

https://pplx.ai/hf-embed-v2-9b
https://pplx.ai/hf-embed-v2-0.6b