[Models](/models)/perplexity-ai/pplx-embed-v2-late-9b

MIT

# pplx-embed-v2-late-9b

by perplexity-ai

Perplexity's 9B late-interaction retriever for text, images and page screenshots, 65.2 nDCG@10 on ViDoRe v3 images

30likes

9B (7.4B active)params

[HuggingFace](https://huggingface.co/perplexity-ai/pplx-embed-v2-late-9b)[Run it on your own data, free](https://studio.mixpeek.com/namespaces/create?mode=managed&utm_source=homepage&utm_medium=model_page&utm_campaign=mvs_onramp)

Identifiers

Model ID

`perplexity-ai/pplx-embed-v2-late-9b`

## Deploy pplx-embed-v2-late-9b

Single-tenant

Mixpeek has no managed extractor for this model. On a single-tenant deployment you upload the weights and a custom plugin serves them next to the rest of your pipeline.

[Read the customer-hosted guide](https://mixpeek.com/docs/customer-hosted/overview) [Talk to us about single-tenant](/demo)

## Overview

pplx-embed-v2-late-9b is the larger of Perplexity's two multimodal late-interaction (ColBERT-style) retrievers, built on Qwen3.5 with bidirectional attention and released under the MIT license. It turns a query or a document into one 128-dimensional vector per token and scores a match with MaxSim, so a query can match the specific words or page regions it is about.

It embeds text, images and visual documents such as page screenshots, but not text and an image in the same input. The card reports 65.2% nDCG@10 on public ViDoRe v3 with page images and 64.7% with markdown, about three points above the 0.6B model, which shares its embedding space.

## Architecture

Qwen3.5 backbone with bidirectional attention, 7.4B active parameters (8.4B in the checkpoint). Outputs one 128-dimensional vector per token and scores query against document with MaxSim. Distilled from an internal 18B ColBERT teacher with a token-level LEAF-style objective; for this model the final eight transformer layers were fully trained.

## Mixpeek SDK Integration

IngestRetrieve

\# No Mixpeek extractor takes a Hugging Face model id, so this model runs on your side.
# It returns one 128-d vector per token; a single-vector index needs them pooled,
# which gives up the token-level matching (see the FAQ below).
from PIL import Image
from sentence\_transformers import MultiVectorEncoder
import requests

model = MultiVectorEncoder("perplexity-ai/pplx-embed-v2-late-9b", device="cuda")
page = Image.open("report-2026-q2-page-14.png").convert("RGB")
token\_vectors = model.encode\_document(\[page\])\[0\]  # (num\_tokens, 128)

requests.post(
    "https://api.mixpeek.com/v1/namespaces/ns\_your\_namespace/documents/upsert",
    headers={"Authorization": "Bearer API\_KEY"},
    json={"collection\_id": "col\_your\_collection",
          "documents": \[{"document\_id": "report-2026-q2-page-14",
                         "vectors": {"page-embedding": token\_vectors.mean(axis=0).tolist()},
                         "payload": {"source\_key": "reports/2026-q2.pdf#page=14"}}\]},
)

## Capabilities

*   One 128-d vector per token, scored with MaxSim (late interaction)
*   Text, image and visual-document inputs, encoded in separate batches
*   Shares an embedding space with pplx-embed-v2-late-0.6b
*   Native Sentence Transformers MultiVectorEncoder, no custom code
*   MIT license

## Use Cases on Mixpeek

Indexing scanned reports, slides and PDFs as page images at the family's best accuracy

Building the index with the 9B and encoding queries with the faster 0.6B

Reranking the top candidates from a cheaper single-vector search

## Benchmarks

Dataset

Metric

Score

Source

Public ViDoRe v3, image

nDCG@10

65.2%

Model card: perplexity-ai/pplx-embed-v2-late-9b (self-reported; 0.6B model 62.3%)

Public ViDoRe v3, markdown

nDCG@10

64.7%

Model card (self-reported; 0.6B model 61.2%)

## Performance

Input SizeText, images or page screenshots; text-only and image-only batches must be encoded separately

Embedding Dim128 per token (multi-vector)

GPU LatencyInput dependent

GPU ThroughputBatch dependent

GPU MemoryModel dependent

Requires sentence-transformers 6.0.0 or later and transformers 5.4.0 or later. Token-level vectors take far more storage than one vector per page. We have not measured it.

## Common Pipeline Companions

[

perplexity-ai/pplx-embed-v2-late-0.6b

The small sibling in the same embedding space, for fast query encoding

](/model/perplexity-ai/pplx-embed-v2-late-0.6b)[

vidore/colqwen2.5-v0.2

An established page-image late-interaction retriever to compare against

](/model/vidore/colqwen2.5-v0.2)

## Frequently Asked Questions

### What is pplx-embed-v2-late-9b?

Perplexity's larger late-interaction retrieval model. It embeds text, images and page screenshots as one 128-dimensional vector per token, scores matches with MaxSim, and is MIT licensed.

### How much better is the 9B than the 0.6B?

On public ViDoRe v3 the card reports 65.2% nDCG@10 with page images and 64.7% with markdown for the 9B, against 62.3% and 61.2% for the 0.6B, about three points, for roughly 20 times the active parameters.

### Can I build the index with the 9B and query with the 0.6B?

Yes. The card says the two models share an embedding space, so you can embed documents once with the 9B and encode queries with the smaller, faster 0.6B.

### Can I store its vectors in a single-vector index?

Only by pooling the per-token vectors into one, for example by averaging, which gives up the token-level matching that makes late interaction accurate. Use the pooled vector for a first pass and rescore the top results with MaxSim over the full token vectors.

## Explore on Mixpeek

[

Browse all models

Search the full model catalog

](/models)[

Feature Extractors

See how Mixpeek runs models as extractors

](/extractors)[

Mixpeek Vector Store

Store & search embeddings at scale

](/mvs)[

Pricing

Usage-based pricing for pipelines

](/pricing)[

All Curated Lists

Compare models, APIs & infrastructure

](/curated-lists)

### Specification

Organizationperplexity-ai

Retriever\-

Parameters9B (7.4B active)

LicenseMIT

Downloads/moN/A

Likes30

[

### Research Paper

Multimodal embeddings beyond a single vector (Perplexity)

arxiv.org](https://www.perplexity.ai/hub/blog/multimodal-embeddings-beyond-a-single-vector)

### Build a pipeline with pplx-embed-v2-late-9b

Add this model to a processing pipeline alongside other extractors. Combine with retrieval stages for end-to-end search.

[Run it on your own data, free](https://studio.mixpeek.com/namespaces/create?mode=managed&utm_source=homepage&utm_medium=model_page&utm_campaign=mvs_onramp)