🔍 Read the full analysis: Explore NVIDIA Nemotron 3.5 Embed, Rerank, And Parse Models on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
NVIDIA has expanded its Nemotron Retriever collection with five models for multimodal embedding, reranking and document parsing. The company reports improved retrieval scores and lower agent inference-token use in selected evaluations, but the supplied announcement lacks full methodology, independent validation and detailed information about Parse.
NVIDIA has expanded its Nemotron Retriever collection with five open models for embedding, reranking and document parsing, targeting AI agents that search text and image-rich documents, as detailed in the original analysis. The company says selected evaluations showed better retrieval results and lower inference-token use, though the figures are company-reported and the supplied announcement does not include independent validation.
The collection includes Nemotron 3.5 Embed 8B BF16, Embed 1B BF16, Embed 1B NVFP4, Rerank 8B BF16 and Parse, a 631-million-parameter model. NVIDIA describes the embedding models as searching across text and visual information, including tables, charts, infographics, datasheets and schematics. The reranker scores query-document pairs to help an agent prioritize candidate evidence.
In evaluations involving ViDoRe v3, BRIGHT and BrowseComp-Plus, NVIDIA says it held the language model, tools, tasks, corpus and execution policy fixed while comparing retrieval setups. It reports that adding Nemotron 3.5 Rerank reduced agent inference tokens per query by about 14% with Embed 8B and 13% with Embed 1B, while improving retrieval accuracy. The company also says Embed 8B scored about 0.9 points higher in nDCG@10 than Qwen3-VL-Reranker-8B at a similar agent token cost.
For ViDoRe v3.1, NVIDIA reports an nDCG@10 score of 61.17 for Embed 8B, compared with 52.23 for Llama Nemotron Embed VL 1B. Embed 1B BF16 scored 58.38, which NVIDIA characterizes as 95.4% of the 8B model’s score. These are results from the company’s stated evaluations, not a guarantee of performance on other datasets or workloads.
Retrieval Can Affect Agent Cost
For an agent handling a large document collection, retrieval determines which evidence reaches the language model. If search returns irrelevant material, the agent may spend more steps inspecting documents or searching again. NVIDIA’s reported results link retrieval quality with task performance and token use, a relationship that can matter to organizations running agents at scale.
The multimodal focus addresses a document-search problem: information needed to answer a question may appear in a chart or table image rather than in text that conventional search indexes easily. Models able to retrieve relevant visual material could help agents work across mixed-format collections. The practical effect remains dependent on the data, workload, hardware and integration; the benchmark figures alone do not establish savings or accuracy gains in a particular enterprise deployment.
NVIDIA Nemotron 3.5 embedding models
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
From Embeddings to Reranking
Embedding models represent queries and documents in a form that lets search systems identify semantically related material. A reranker then compares a query with candidate results and adjusts their order. NVIDIA’s described approach puts both stages within an agent workflow that can search, inspect evidence and search again if needed.
The model variants offer different options: NVIDIA positions Embed 8B BF16 for quality-focused retrieval and Embed 1B BF16 as a smaller multimodal option. Embed 1B NVFP4 uses quantization for text and image encoders, which NVIDIA presents as an inference-efficiency option on supported hardware. The company also says its fine-tuning recipes can adapt embeddings to custom collections. In chart and infographic experiments, it reports relative nDCG@10 gains of 18% for Embed 1B BF16 and 3% for Embed 8B BF16.
NVIDIA says it evaluated text retrieval across 34 languages, and multilingual visual retrieval on MIRACL Vision across 18 languages and on KoViDoRe. Those figures describe the reported evaluation scope; the announcement’s suggestion that support may extend further does not establish results in additional languages. The new collection extends the company’s earlier Nemotron 3 Embed text embedding models.
“Adding Nemotron 3.5 Rerank reduces agentic token cost per query by approximately 14% with Nemotron 3.5 Embed 8B, and 13% with Nemotron 3.5 Embed 1B, while improving retrieval accuracy in both configurations.”
— NVIDIA, in its Hugging Face announcement
As an affiliate, we earn on qualifying purchases.
Benchmark and Parse Details
The supplied announcement does not provide full benchmark methodology or results tables, and it includes no independent replication of the reported scores or token savings. Results may vary across document collections, languages, hardware and agent tasks, so the figures should be read within the company’s stated evaluation setup.
Information about Nemotron 3.5 Parse is particularly limited: the source names the model and its parameter count but does not detail its capabilities, supported formats, evaluation or relationship to the embedding and reranking workflow. The material also does not specify licensing terms, pricing, complete hardware requirements, release links or a broader availability schedule. It describes the models as open and commercially available, but the supplied information does not clarify those release terms.
multimodal embedding models for AI
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Documentation and Outside Tests
Readers and prospective users will need the model documentation and release artifacts to check licensing, access, deployment requirements and supported hardware. Those materials could also clarify how Parse fits into the collection and provide the detail needed to reproduce NVIDIA’s evaluations.
Independent testing on enterprise document collections would show whether the reported retrieval gains and token reductions carry over to different data and agent workflows. Until such results and fuller release details are available, the announcement establishes NVIDIA’s reported benchmarks and product lineup, but not their performance or cost impact in a specific deployment.
As an affiliate, we earn on qualifying purchases.
Key Questions
Which models did NVIDIA add?
The collection includes Embed 8B BF16, Embed 1B BF16, Embed 1B NVFP4, Rerank 8B BF16 and Parse. NVIDIA lists Parse as a 631-million-parameter model.
What retrieval gains did NVIDIA report?
NVIDIA says adding Nemotron 3.5 Rerank reduced agent inference tokens per query by about 14% with Embed 8B and 13% with Embed 1B, while improving retrieval accuracy in its selected evaluations.
Have the benchmark results been independently verified?
The supplied announcement provides company-reported results but no independent replication. It also does not include complete methodology or results tables.
What is known about Nemotron 3.5 Parse?
The source identifies Parse as a 631-million-parameter model, but does not give full details about its capabilities, supported document formats, evaluation or availability.
Are the models available, and what are the terms?
The announcement describes the models as open and commercially available, but the supplied material does not include release links, licensing details, pricing or a broader availability schedule.
Primary source: Hugging Face · via ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
