NeMo Retriever

Type: Platform Tags: NVIDIA, NeMo Retriever, RAG, retrieval, embedding, reranking, multimodal data extraction Related: NVIDIA-NeMo, NeMo-Platform, NVIDIA-NIM, NVIDIA-NIM-Operator, NVIDIA-RAG-Blueprint, NeMo-Retriever-Embedding-NIM, Llama-Nemotron-Embed-1B-v2, Llama-Nemotron-Embed-VL-1B-v2, NIM-for-NV-CLIP, NeMo-Retriever-Reranking-NIM, Llama-Nemotron-Rerank-1B-v2, Llama-Nemotron-Rerank-VL-1B-v2, NIM-for-Image-OCR, NIM-for-Object-Detection, Nemotron-Parse, NVIDIA-AI-Data-Platform, NVIDIA-AI-Q-Blueprint, NVIDIA-AI-Blueprints, cuVS, NVIDIA-Agent-Intelligence-Toolkit, Nemotron Sources: https://docs.nvidia.com/nemo/retriever/latest/index.html, https://docs.nvidia.com/rag/latest/, https://docs.nvidia.com/nim/nemo-retriever/text-embedding/latest/overview.html, https://docs.nvidia.com/nim/nvclip/latest/introduction.html, https://docs.nvidia.com/nim/nemo-retriever/text-reranking/latest/overview.html, https://docs.nvidia.com/nim/ingestion/image-ocr/latest/overview.html, https://docs.nvidia.com/nim/ingestion/object-detection/latest/overview.html, https://docs.nvidia.com/nim/vision-language-models/latest/examples/nemotron-parse/api.html, https://www.nvidia.com/en-us/data-center/ai-data-platform/, https://docs.nvidia.com/aiq-blueprint/latest/index.html Last Updated: 2026-04-29

Summary

NeMo Retriever is NVIDIA’s collection of microservices for building and scaling retrieval pipelines with multimodal data extraction, embeddings, indexing, retrieval, and reranking. It is built with NVIDIA-NIM and is part of the NeMo software suite for AI agent lifecycle management.

Detail

Purpose

Enterprise RAG and agent systems need to connect models to proprietary data with privacy, accuracy, and scale. NeMo Retriever provides document extraction, embedding, indexing, semantic/hybrid search, and reranking services optimized for NVIDIA infrastructure.

Key capabilities

• Multimodal data extraction for documents with text, tables, charts, and infographics, including NIM-for-Image-OCR, NIM-for-Object-Detection, and adjacent VLM-style parsing with Nemotron-Parse.
• Embedding and indexing flows using vector databases such as LanceDB or Milvus, including NeMo-Retriever-Embedding-NIM.
• Multimodal text/image embedding and semantic image retrieval through NIM-for-NV-CLIP.
• GPU-accelerated vector search and indexing through cuVS.
• Semantic and hybrid retrieval with embedding and reranking NIM microservices such as NeMo-Retriever-Embedding-NIM and NeMo-Retriever-Reranking-NIM.
• Model-specific text retrieval using Llama-Nemotron-Embed-1B-v2 for candidate generation and Llama-Nemotron-Rerank-1B-v2 for passage reranking.
• Model-specific multimodal document retrieval using Llama-Nemotron-Embed-VL-1B-v2 for candidate generation and Llama-Nemotron-Rerank-VL-1B-v2 for visual-document reranking.
• Multimodal visual-document reranking with model-specific endpoints such as Llama-Nemotron-Rerank-VL-1B-v2.
• Enterprise privacy and deployment model consistent with self-hosted NVIDIA AI stacks.

NVIDIA context

NeMo Retriever is a central bridge between enterprise data and NVIDIA agent systems. It connects NVIDIA-NIM model endpoints, NVIDIA-Agent-Intelligence-Toolkit workflows, Nemotron reasoning models, NVIDIA-AI-Data-Platform reference workflows, NVIDIA-RAG-Blueprint deployments, and vector-search acceleration.

Connections

• NVIDIA-NeMo - Retriever is part of the NeMo suite.
• NeMo-Platform - platform-level deployment and lifecycle management for NeMo services.
• NVIDIA-NIM - Retriever components are delivered as NIM-style microservices.
• NVIDIA-NIM-Operator - Kubernetes deployment and lifecycle management path for NIM/NeMo microservices.
• NVIDIA-RAG-Blueprint - reference RAG workflow that composes Retriever-style extraction, embedding, retrieval, and generation.
• NeMo-Retriever-Embedding-NIM - embedding microservice for semantic search and RAG.
• Llama-Nemotron-Embed-1B-v2 - NVIDIA text embedding model for multilingual long-document QA retrieval.
• Llama-Nemotron-Embed-VL-1B-v2 - NVIDIA multimodal embedding model for visual document/page retrieval.
• NIM-for-NV-CLIP - multimodal text/image embedding NIM for semantic image search and multimodal RAG.
• NeMo-Retriever-Reranking-NIM - reranking microservice for improving candidate context relevance.
• Llama-Nemotron-Rerank-1B-v2 - NVIDIA text reranker model for improving retrieved passage order.
• Llama-Nemotron-Rerank-VL-1B-v2 - NVIDIA multimodal reranker model for pages, slides, text, and image+text documents.
• NIM-for-Image-OCR - OCR extraction microservice for images and visual document regions.
• NIM-for-Object-Detection - document object-detection microservices for page, table, and graphic elements.
• Nemotron-Parse - VLM document parser for text/table extraction, semantic classes, and bounding boxes.
• NVIDIA-AI-Data-Platform - AI Data Platform uses retrieval, indexing, and vector search to connect agents to enterprise data.
• NVIDIA-AI-Q-Blueprint - AI-Q uses retrieval and knowledge sources as part of its enterprise research workflow.
• NVIDIA-AI-Blueprints - RAG, deep research, and data search blueprints are discovery surfaces for Retriever-related patterns.
• cuVS - GPU vector search acceleration used in indexing/retrieval stacks.
• NVIDIA-Agent-Intelligence-Toolkit - agent workflows use retrievers to access enterprise data.

Source Excerpts

• NVIDIA NeMo Retriever docs describe multimodal extraction, embedding/indexing, retrieval, and reranking microservices.