AI news ·
NVIDIA Releases Nemotron 3.5 Retriever Models
NVIDIA released Nemotron 3.5 Retriever models for multimodal search and document parsing, but gave no pricing, dates, or benchmarks.
NVIDIA has released Nemotron 3.5 Retriever models, a set of tools for multimodal search, document parsing, and reranking. The models are designed to support retrieval infrastructure that feeds grounded assistants, where the quality of retrieved context directly affects response accuracy.
Organizations evaluating the models should test them on their own documents. NVIDIA notes that buyers need to validate domain accuracy, latency, and provenance independently. The release targets teams working across fields such as healthcare, insurance, IT and development, product development, science and research, education, and creative industries.
The models address a technical gap in retrieval pipelines by improving how systems locate and rank information across different data formats before it reaches a language model. No pricing, availability dates, or performance benchmarks were included in the announcement.