120B / 12B active (LatentMoE)
NVIDIA Nemotron Open Model License
Open Nemotron 3 Super release family with base and post-trained checkpoints plus optimized precision variants (BF16, FP8, NVFP4); hybrid LatentMoE architecture combining interleaved Mamba-2, MoE, and attention layers with Multi-Token Prediction (MTP); up to 1M-token context window and reasoning mode controls; release positioning emphasizes agentic workflows, tool use, long-context reasoning, and high-throughput enterprise workloads, alongside open dataset collections for pre-training and post-training.
https://research.nvidia.com/labs/nemotron/files/NVIDIA-Nemotron-3-Super-Technical-Report.pdf