NVIDIA
NVIDIA is best known as the GPU company that powers the AI industry, but their model lineup matters too. The Nemotron family is purpose-built to showcase what their hardware can do, and the April 2026 release of Nemotron 3 Nano Omni 30B-A3B-Reasoning landed as one of the strongest open multimodal models of the year. It processes text, image, video, and audio in a single unified sequence via a hybrid Mamba-Transformer-MoE backbone (30B total parameters, 3B active per token), with a 256K token context window and native audio handling up to 20 minutes per clip. It tops six public leaderboards for document intelligence, video understanding, and voice interaction, and is available on Hugging Face under an open weight license in BF16, FP8, and NVFP4 quantizations including consumer-GPU formats.
Founded
1993
Headquarters
Santa Clara, CA
CEO
Jensen Huang
Models
1 active
Key Products
Strengths
- ✓Open weights with consumer GPU support
- ✓256K context multimodal
- ✓Top document and video benchmarks
- ✓Native audio as first-class modality
- ✓Self-hosted deployment focus
NVIDIA Models
| Model | Input / 1M | Output / 1M | Context | Capabilities |
|---|---|---|---|---|
| Nemotron 3 Nano Omni | Free | Free | 256K | text, vision, audio, video, code, reasoning, tool-use |
Prices per 1M tokens in USD. See the full pricing guide for detailed analysis.