nvidia/Nemotron-3.5-Embed-1B-BF16 · Hugging Face
Nemotron-3.5-Embed-1B-BF16 is a versatile embedding model developed by NVIDIA for multimodal and multilingual question-answering retrieval.
NVIDIAHugging FaceRetrieval-augmented generationnvidia/Nemotron-3.5-Embed-8B-BF16nvidia/Nemotron-3.5-Embed-1B-BF16
Proof1 community thread
Key points
- The model can embed a document page in the form of an image, text, or combined image–text input.
- For text retrieval, the model was evaluated across 34 languages: English, Arabic, Assamese, Bengali, Bulgarian, Chinese, Danish, Dutch, Finnish, French, German, Hindi, Hinglish, Indonesian, Italian, Japanese, Korean, Malay, Marathi, Nepali, Norwegian, Persian, Portuguese, Romanian, Russian, Spanish, Swahili, Swedish, Tamil, Telugu, Thai, Ukrainian, Urdu, and Vietnamese.
- The embedding model is a bi-encoder that supports context in textual format (e.g. the query or the OCR text of a page or a section of a document) or the image of a document page.
- The model generates dense vector embeddings from text or image inputs, enabling multimodal retrieval, semantic search, and (agentic) RAG workflows.
Sources (1)
- [1]nvidia/Nemotron-3.5-Embed-1B-BF16 · Hugging Facer/LocalLLaMA (top, daily) · Oct 15, 11:22 PM
Nemotron-3.5-Embed-1B-BF16 is a versatile embedding model developed by NVIDIA for multimodal and multilingual question-answering retrieval.
The model can embed a document page in the form of an image, text, or combined image–text input.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 15, 2026Granite Switch 4.1: Chat, Decisions, and Speech from One Endpoint
- Oct 15, 2026unslothai/unsloth v0.1.908-beta: NVIDIA Nemotron 3.5 + Meta Muse Spark Open
- Oct 15, 2026NVIDIA Nemotron 3.5 Super VL Powers Specialized Agents with Multimodal Reasoning
- Oct 15, 2026nvidia/Nemotron-3.5-Rerank-8B-BF16
- Oct 14, 2026Deploy and evaluate models with Amazon SageMaker JumpStart
- Oct 14, 2026mistralai/Timestral-209M
