Back to Project
N

Nemotron 3.5 retrieval models

Project Oct 21, 2026

NVIDIA's October 2026 family of multimodal embedding, reranking and document parsing models for retrieval systems and AI agents.

Visit primary link 2 sources attached
NVIDIA's scatter plot compares average retrieval accuracy with agentic token cost per query across embedding and reranking configurations.
NVIDIA-run comparison on ViDoRe v3, BRIGHT and BrowseComp-Plus, holding the agent setup fixed. The horizontal axis measures agent token cost, not the complete cost of retrieval and infrastructure. These are vendor results, not an AITrending reproduction. Source: NVIDIA
multimodalai-agentsself-hosted

What happened lately

Nemotron 3.5 retrieval models timeline

The family includes 8B and 1B embedding models, a quantized 1B variant, an 8B reranker and a 631M document parser. Its components target visual-document search, relevance scoring and structured extraction. The Embed 8B card documents an OpenMDW-1.1 license and Linux GPU inference examples. Component licenses and runtime requirements need checking individually.