NVIDIA Nemotron 3 Embed is a collection of open embedding models for retrieval tasks. Nemotron-3-Embed-8B-BF16 ranks #1 on the RTEB leaderboard, scoring 78.5%. The collection includes 1B variants for efficient deployment and features a 32k context window, multilingual support, and NVFP4 quantization for Blackwell GPUs. NVIDIA reports the 8B model achieves the lowest estimated downstream agentic token cost across multiple benchmarks.
No score is assigned. Sources and their independence are shown in the citation chain below.