NVIDIA
NVIDIA develops the Nemotron model family alongside its computing hardware and software. The family includes models trained by NVIDIA and models adapted from other developers' releases. The entries identify the reviewed weights and their specific purpose.
Models & tools
- Nemotron 3.5 Super VL 120B-A12B
NVIDIA · Nemotron 3.5
A 120B-total, 12B-active vision-language model that combines a hybrid language backbone with a C-RADIO encoder.
Launch date not confirmedDownloadable - Nemotron 3 Ultra 550B-A55B
NVIDIA · Nemotron 3
A hybrid LatentMoE reasoning model with 550B total and 55B active parameters.
Launch date not confirmedDownloadable - Nemotron 3.5 Lightning 30B-A3B
NVIDIA · Nemotron 3.5
A hybrid text model with Mamba-2, expert and attention layers, 30B total parameters and 3B active parameters.
Launch date not confirmedDownloadable - Nemotron 3 Nano Omni 30B-A3B Reasoning
NVIDIA · Nemotron 3
A multimodal text-output model with 31B total parameters that accepts video, audio, images and text.
Launch date not confirmedDownloadable - Nemotron 3 Super 120B-A12B
NVIDIA · Nemotron 3
A text reasoning model with 120B total and 12B active parameters.
Launch date not confirmedDownloadable - Nemotron 3 Nano 30B-A3B
NVIDIA · Nemotron 3
A text model with hybrid Mamba-2, expert and attention layers and 3.5B active parameters.
Launch date not confirmedDownloadable - Nemotron 3.5 Rerank 8B
NVIDIA · Nemotron 3.5
An 8.38B relevance model that scores text queries against text, image or combined documents.
Launch date not confirmedDownloadable - Nemotron 3.5 Content Safety
NVIDIA · Nemotron 3.5
A 4B Gemma-derived classifier for multimodal safety assessment and custom-policy reasoning.
Launch date not confirmedDownloadable - Nemotron 3.5 Embed 1B
NVIDIA · Nemotron 3.5
A 1.55B embedding model that maps text queries and text or image documents into retrieval vectors.
Launch date not confirmedDownloadable - Nemotron 3.5 Embed 8B
NVIDIA · Nemotron 3.5
An 8.38B embedding model for text queries and documents containing text, images or both.
Launch date not confirmedDownloadable - Nemotron 3.5 ASR Streaming 0.6B
NVIDIA · Nemotron 3.5
A 600M speech recognizer for streaming transcription across 40 language-locales.
Launch date not confirmedDownloadable - Nemotron 3.5 Transcribe 30B-A4B
NVIDIA · Nemotron 3.5
An offline audio-to-text model for 40 languages that accepts recordings, instructions and optional context.
Launch date not confirmedDownloadable - Nemotron 3 Nano 4B
NVIDIA · Nemotron 3
A compact text model compressed from Nemotron Nano 9B v2 using NVIDIA's Elastic method.
Launch date not confirmedDownloadable - Nemotron 3.5 Bolt 74B-A9B
NVIDIA · Nemotron 3.5
A compressed Nemotron 3.5 Super VL derivative with 74B total and 9B active parameters.
Launch date not confirmedDownloadable - Nemotron 3.5 Bolt E4B
NVIDIA · Nemotron 3.5
A text model produced by NVIDIA's further post-training of Google's Gemma 4 E4B instruction model.
Launch date not confirmedDownloadable - Nemotron 3.5 Parse
NVIDIA · Nemotron 3.5
A document model that converts page images into text, tables, chart data, reading order and bounding boxes.
Launch date not confirmedDownloadable