Skip to main content
BenchLM
Data

ViDoRe V3 Markdown (Perplexity report)

We show this table for reference; we do not rank on it.

Perplexity reports public ViDoRe V3 Markdown retrieval nDCG@10 of 61.2 for the 0.6B model and 64.7 for the 9B model.

nDCG@10 on ViDoRe V3 Markdown (Perplexity report) — October 7, 2026 model-card report

We mirror the published ndcg@10 view for ViDoRe V3 Markdown (Perplexity report). The public snapshot contains 2 models. pplx-embed-v2-late-9b has the highest published score at 64.7%, but the available coverage and evaluation setups do not establish a market leader. We do not use these results to rank models overall.

1Open weights

pplx-embed-v2-late-9b

Perplexity

Public ViDoRe V3; Markdown

64.7%
Context 4096
2Open weights

pplx-embed-v2-late-0.6b

Perplexity

Public ViDoRe V3; Markdown

61.2%
Context 4096

2 modelsMultimodal & GroundedCurrentDisplay onlyUpdated October 7, 2026 model-card report

nDCG@10 table (2 models)

nDCG@10 results for ViDoRe V3 Markdown (Perplexity report)
RankModel / configurationScoreParameters (B)Open / closed
1pplx-embed-v2-late-9bPerplexityPublic ViDoRe V3; Markdown
64.7%
Not reportedOpen
2pplx-embed-v2-late-0.6bPerplexityPublic ViDoRe V3; Markdown
61.2%
Not reportedOpen

Public ViDoRe V3 retrieval results

Perplexity reports public ViDoRe V3 nDCG@10 on a 0–100 scale. Image and Markdown retrieval use separate input representations. These provider-reported results remain display-only and do not establish general LLM capability scores. The two models share a multi-vector embedding space; each retained token has a 128-dimensional vector, and query–document pairs use MaxSim scoring.

Snapshot

Public ViDoRe V3Provider-reportedDisplay only
Perplexity multimodal embedding launch

About ViDoRe V3 Markdown (Perplexity report)

Year

2026

Format

Public ViDoRe V3 retrieval

Difficulty

Visual document retrieval

Perplexity reports public ViDoRe V3 nDCG@10 on a 0–100 scale. Image and Markdown retrieval use separate input representations. These provider-reported results remain display-only and do not establish general LLM capability scores. The two models share a multi-vector embedding space; each retained token has a 128-dimensional vector, and query–document pairs use MaxSim scoring.

Perplexity multimodal late-interaction embedding model cardPublic benchmark source

Freshness and provenance

Version

ViDoRe V3 Markdown (Perplexity report) 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

Questions

What does ViDoRe V3 Markdown (Perplexity report) measure?

Perplexity reports public ViDoRe V3 Markdown retrieval nDCG@10 of 61.2 for the 0.6B model and 64.7 for the 9B model.

Which model leads the published ViDoRe V3 Markdown (Perplexity report) snapshot?

pplx-embed-v2-late-9b currently leads the published ViDoRe V3 Markdown (Perplexity report) snapshot with 64.7% ndcg@10. BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on ViDoRe V3 Markdown (Perplexity report)?

The October 7, 2026 model-card report snapshot contains 2 AI models.

Last updated: October 7, 2026 model-card report · mirrored from the public benchmark leaderboard

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 5,600+ readers.

One email each week. Unsubscribe anytime.