Nemotron 3 Super: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning

• URL: http://arxiv.org/abs/2604.12374v1
• Date: Tue, 14 Apr 2026 07:02:32 GMT
• Title: Nemotron 3 Super: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning
• Abstract summary: We describe the pre-training, post-training, and quantization of Nemotron 3 Super.<n>Nemotron 3 Super is a 120 billion (active 12 billion) parameter hybrid Mamba-Attention Mixture-of-Experts model.
• Score: 262.0414794728708
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: We describe the pre-training, post-training, and quantization of Nemotron 3 Super, a 120 billion (active 12 billion) parameter hybrid Mamba-Attention Mixture-of-Experts model. Nemotron 3 Super is the first model in the Nemotron 3 family to 1) be pre-trained in NVFP4, 2) leverage LatentMoE, a new Mixture-of-Experts architecture that optimizes for both accuracy per FLOP and accuracy per parameter, and 3) include MTP layers for inference acceleration through native speculative decoding. We pre-trained Nemotron 3 Super on 25 trillion tokens followed by post-training using supervised fine tuning (SFT) and reinforcement learning (RL). The final model supports up to 1M context length and achieves comparable accuracy on common benchmarks, while also achieving up to 2.2x and 7.5x higher inference throughput compared to GPT-OSS-120B and Qwen3.5-122B, respectively. Nemotron 3 Super datasets, along with the base, post-trained, and quantized checkpoints, are open-sourced on HuggingFace.

Related papers

• Tevatron Meets Megatron: Expert-Parallel LLM Reranker Training on an Academic Budget [72.7352550413971]
  We present Tevatron 3.0, which integrates a Megatron-Core training backend into Tevatron.<n>We show that Megatron matches FSDP reranker quality and training efficiency under comparable data-parallel settings.<n>We find that the MoE reranker matches dense 8B quality while activating less than half as many parameters and achieving substantially higher inference throughput.
  arXiv  Detail & Related papers  (2026-08-02T00:55:47Z)
• Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning [269.10652784788374]
  Nemotron 3 Ultra is a 550 billion total and 55 billion active parameter Mixture-of-Experts Hybrid Mamba-Attention language model.<n>We pre-trained Nemotron 3 Ultra on 20 trillion text tokens, then extended the context length to 1M tokens, and post-trained using Supervised Fine Tuning (SFT), Reinforcement Learning (RL), and Multi-teacher On-Policy Distillation (MOPD)
  arXiv  Detail & Related papers  (2026-06-12T22:56:12Z)
• NVIDIA Nemotron 3: Efficient and Open Intelligence [227.47413816066845]
  Nemotron 3 family of models deliver strong agentic, reasoning, and conversational capabilities.<n>Nemotron 3 models are post-trained using multi-environment reinforcement learning enabling reasoning, multi-step tool use, and support granular reasoning budget control.<n>Nemotron 3 family uses a Mixture-of-Experts hybrid Mamba-Transformer architecture to provide best-in-class throughput and context lengths of up to 1M tokens.
  arXiv  Detail & Related papers  (2025-12-24T00:24:05Z)
• Nemotron 3 Nano: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning [223.88531320605486]
  Nemotron 3 Nano 30B-A3B is a Mixture-of-Experts hybrid Mamba-Transformer language model.<n>Nemotron 3 Nano was pretrained on 25 trillion text tokens, including more than 3 trillion new unique tokens over Nemotron 2.
  arXiv  Detail & Related papers  (2025-12-23T23:54:32Z)
• Jet-Nemotron: Efficient Language Model with Post Neural Architecture Search [42.46046429414803]
  Jet-Nemotron is a new family of hybrid-architecture language models.<n>It matches or exceeds the accuracy of leading full-attention models.
  arXiv  Detail & Related papers  (2025-08-21T17:59:08Z)
• NVIDIA Nemotron Nano 2: An Accurate and Efficient Hybrid Mamba-Transformer Reasoning Model [205.51647932840226]
  Nemotron-Nano-9B-v2 is a hybrid Mamba-Transformer language model designed to increase throughput for reasoning workloads.<n>Nemotron-Nano-9B-v2 builds on the Nemotron-H architecture, in which the majority of the self-attention layers in the common Transformer architecture are replaced with Mamba-2 layers.
  arXiv  Detail & Related papers  (2025-08-20T06:00:57Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.