NVIDIA Releases Nemotron-3 Embedding Model

nvidia · hf · 2026-07-14

NVIDIA has released **Nemotron-3-Embed-1B-BF16**, a new **feature-extraction / text-embeddings** model designed for retrieval, semantic search, and RAG scenarios. The post also provides implementation and deployment details, noting its multilingual support, BF16 format, and safetensors compatibility.

Related event: NVIDIA Launches Nemotron-3-Embed Models(2 posts)→

Original post →

More from Infra

Infra channel →