mlsanityData to 14 Oct 2026

NVIDIA Nemotron Nano 2: An Accurate and Efficient Hybrid Mamba-Transformer Reasoning Model

arXiv 2508.14444 ↗First posted 20 Aug 2025cs.CL cs.AI cs.LGcited by 108

On the map: LLM reasoning → Test-time compute and efficient thinking

also in 1 more
Abstract (arXiv)

We introduce Nemotron-Nano-9B-v2, a hybrid Mamba-Transformer language model designed to increase throughput for reasoning workloads while achieving state-of-the-art accuracy compared to similarly-sized models. 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, to achieve improved inference speed when generating the long thinking traces needed for reasoning. We create Nemotron-Nano-9B-v2 by first pre-training a 12-billion-parameter model (Nemotron-Nano-12B-v2-Base) on 20 trillion tokens using an FP8 training recipe. After aligning Nemotron-Nano-12B-v2-Base, we employ the Minitron strategy to compress and distill the model with the goal of enabling inference on up to 128k tokens on a single NVIDIA A10G GPU (22GiB of memory, bfloat16 precision). Compared to existing similarly-sized models (e.g., Qwen3-8B), we show that Nemotron-Nano-9B-v2 achieves on-par or better accuracy on reasoning benchmarks while achieving up to 6x higher inference throughput in reasoning settings like 8k input and 16k output tokens. We are releasing Nemotron-Nano-9B-v2, Nemotron-Nano12B-v2-Base, and Nemotron-Nano-9B-v2-Base checkpoints along with the majority of our pre- and post-training datasets on Hugging Face.

Key facts. Cited by 108 papers in the map (data through 14 Oct 2026). Placed in Test-time compute and efficient thinking, one of 17 directions in LLM reasoning. Builds on 59 of its 76 references that are in the map. First posted on arXiv 20 Aug 2025.

Builds on

59 of 76 references.

Closest ancestors · 2

Same direction, newest first

  1. 2025Llama-Nemotron: Efficient Reasoning Models2 May
  2. 2025Nemotron-H: A Family of Accurate and Efficient Hybrid Mamba-Transformer Models4 Apr

Other references · 57

  1. 2022Training language models to follow instructions with human feedbackcited by 15,971
  2. 2021Training Verifiers to Solve Math Word Problemscited by 10,109
  3. 2024DeepSeekMath: Pushing the Limits of Mathematical Reasoning in Open Language Modelscited by 9,379
  4. 2023Judging LLM-as-a-Judge with MT-Bench and Chatbot Arenacited by 8,626
  5. 2021Evaluating Large Language Models Trained on Codecited by 8,320
  6. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  7. 2020Measuring Massive Multitask Language Understandingcited by 7,963
  8. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
  9. 2021Measuring Mathematical Problem Solving With the MATH Datasetcited by 5,371
  10. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,886
  11. 2019HellaSwag: Can a Machine Really Finish Your Sentence?cited by 4,100
  12. 2017Deep reinforcement learning from human preferencescited by 3,912
38 more
  1. 2023Let's Verify Step by Stepcited by 3,789
  2. 2021Program Synthesis with Large Language Modelscited by 3,637
  3. 2019PIQA: Reasoning about Physical Commonsense in Natural Languagecited by 2,883
  4. 2018Can a Suit of Armor Conduct Electricity? A New Dataset for Open Book Question Answeringcited by 2,164
  5. 2024LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Codecited by 2,004
  6. 2019Root Mean Square Layer Normalizationcited by 1,480
  7. 2022Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference timecited by 1,396
  8. 2023Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generationcited by 1,391
  9. 2023GQA: Training Generalized Multi-Query Transformer Models from Multi-Head Checkpointscited by 1,346
  10. 2024RULER: What's the Real Context Size of Your Long-Context Language Models?cited by 1,346
  11. 2024Transformers are SSMs: Generalized Models and Efficient Algorithms Through Structured State Space Dualitycited by 1,336
  12. 2022Text Embeddings by Weakly-Supervised Contrastive Pre-trainingcited by 1,253
  13. 2023Instruction-Following Evaluation for Large Language Modelscited by 1,228
  14. 2024MiniCPM: Unveiling the Potential of Small Language Models with Scalable Training Strategiescited by 723
  15. 2024Phi-4 Technical Reportcited by 711
  16. 2017Program Induction by Rationale Generation : Learning to Solve and Explain Algebraic Word Problemscited by 650
  17. 2024WildChat: 1M ChatGPT Interaction Logs in the Wildcited by 629
  18. 2022Language Models are Multilingual Chain-of-Thought Reasonerscited by 530
  19. 2023Physics of Language Models: Part 3.1, Knowledge Storage and Extractioncited by 270
  20. 2024JailBreakV: A Benchmark for Assessing the Robustness of MultiModal Large Language Models against Jailbreak Attackscited by 230
  21. 2020Retrieve, Program, Repeat: Complex Knowledge Base Question Answering via Alternate Meta-learningcited by 216
  22. 2024ToolACE: Winning the Points of LLM Function Callingcited by 216
  23. 2024OpenMathInstruct-2: Accelerating AI for Math with Massive Open-Source Instruction Datacited by 200
  24. 2023Rethinking Benchmark and Contamination for Language Models with Rephrased Samplescited by 195
  25. 2024Global MMLU: Understanding and Addressing Cultural and Linguistic Biases in Multilingual Evaluationcited by 191
  26. 2024An Empirical Study of Mamba-based Language Modelscited by 186
  27. 2025AIMO-2 Winning Solution: Building State-of-the-Art Mathematical Reasoning Models with OpenMathReasoning datasetcited by 174
  28. 2025APIGen-MT: Agentic Pipeline for Multi-Turn Data Generation via Simulated Agent-Human Interplaycited by 167
  29. 2021Primer: Searching for Efficient Transformers for Language Modelingcited by 164
  30. 2024Nemotron-CC: Transforming Common Crawl into a Refined Long-Horizon Pretraining Datasetcited by 160
  31. 2025FineWeb2: One Pipeline to Scale Them All -- Adapting Pre-Training Data Processing to Every Languagecited by 143
  32. 2025Aegis2.0: A Diverse AI Safety Dataset and Risks Taxonomy for Alignment of LLM Guardrailscited by 132
  33. 2025OpenCodeReasoning: Advancing Data Distillation for Competitive Codingcited by 113
  34. 2025HelpSteer3-Preference: Open Human-Annotated Preference Data across Diverse Tasks and Languagescited by 75
  35. 2025OpenCodeInstruct: A Large-scale Instruction Tuning Dataset for Code LLMscited by 60
  36. 2024WorkBench: a Benchmark Dataset for Agents in a Realistic Workplace Settingcited by 53
  37. 2025MegaMath: Pushing the Limits of Open Math Corporacited by 51
  38. 2021From LSAT: The Progress and Challenges of Complex Reasoningcited by 43

50 most cited of 57 All 57 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 25; 2026: 83.

108 papers.

Sort:

By area:

  1. 2025Nemotron 3 Nano: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoningcited by 104
  2. 2025MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scalingcited by 81
  3. 2025Pretraining Large Language Models with NVFP4cited by 76
  4. 2025NVIDIA Nemotron 3: Efficient and Open Intelligencecited by 65
  5. 2025Nemotron-Cascade: Scaling Cascaded Reinforcement Learning for General-Purpose Reasoning Modelscited by 48
  6. 2025Teaching Pretrained Language Models to Think Deeper with Retrofitted Recurrencecited by 47
  7. 2025HAMLET: Switch your Vision-Language-Action Model into a History-Aware Policycited by 46
  8. 2026Nemotron 3 Super: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoningcited by 38
  9. 2025Efficient-DLM: From Autoregressive to Diffusion Language Models, and Beyond in Speedcited by 33
  10. 2025Front-Loading Reasoning: The Synergy between Pretraining and Post-Training Datacited by 27
  11. 2026Scaling Beyond Masked Diffusion Language Modelscited by 27
  12. 2025RLP: Reinforcement as a Pretraining Objectivecited by 25
38 more
  1. 2026Nemotron-Labs-Diffusion: A Tri-Mode Language Model Unifying Autoregressive, Diffusion, and Self-Speculation Decodingcited by 24
  2. 2026Quantization-Aware Distillation for NVFP4 Inference Accuracy Recoverycited by 22
  3. 2026Olmo Hybrid: From Theory to Practice and Backcited by 22
  4. 2025Don't Pass@k: A Bayesian Framework for Large Language Model Evaluationcited by 15
  5. 2025Nemotron-Flash: Towards Latency-Optimal Hybrid Small Language Modelscited by 15
  6. 2025Hybrid Architectures for Language Models: Systematic Analysis and Design Insightscited by 14
  7. 2026HEARTS: Benchmarking LLM Reasoning on Health Time Seriescited by 11
  8. 2026Data-Efficient Autoregressive-to-Diffusion Language Models via On-Policy Distillationcited by 11
  9. 2026Where Do the Joules Go? Diagnosing Inference Energy Consumptioncited by 10
  10. 2026HySparse: A Hybrid Sparse Attention Architecture with Oracle Token Selection and KV Cache Sharingcited by 10
  11. 2026DFlare: Scaling Up Draft Capacity for Block Diffusion Speculative Decodingcited by 9
  12. 2025Hermes 4 Technical Reportcited by 8
  13. 2026Lost in the Noise: How Reasoning Models Fail with Contextual Distractorscited by 8
  14. 2026MolecularIQ: Characterizing Chemical Reasoning Capabilities Through Symbolic Verification on Molecular Graphscited by 8
  15. 2026PARD-2: Target-Aligned Parallel Draft Model for Dual-Mode Speculative Decodingcited by 8
  16. 2025Are Large Reasoning Models Interruptible?cited by 7
  17. 2025Nemotron Elastic: Towards Efficient Many-in-One Reasoning LLMscited by 6
  18. 2025Where Did This Sentence Come From? Tracing Provenance in LLM Reasoning Distillationcited by 6
  19. 2026Unlocking Reasoning Capability on Machine Translation in Large Language Modelscited by 6
  20. 2026Don't Retrain, Align: Adapting Autoregressive LMs to Diffusion LMs via Representation Alignmentcited by 6
  21. 2026FLARE: Diffusion for Hybrid Language Modelcited by 6
  22. 2026Test-Time Scaling in Reasoning LLMs: Inference Regimes, Evaluation, and Reproducibilitycited by 6
  23. 2025IndustryNav: Exploring Spatial Reasoning of Embodied Agents in Dynamic Industrial Navigationcited by 5
  24. 2026Priming: Hybrid State Space Models From Pre-trained Transformerscited by 5
  25. 2026Agentic Discovery of Neural Architectures: AIRA-Compose and AIRA-Designcited by 5
  26. 2025TimeViper: A Hybrid Mamba-Transformer Vision-Language Model for Efficient Long Video Understandingcited by 4
  27. 2026Bielik-Minitron-7B: Compressing Large Language Models via Structured Pruning and Knowledge Distillation for the Polish Languagecited by 4
  28. 2026Can LLM Agents Be CFOs? Benchmarking Long-Horizon Resource Allocation in an Uncertain Enterprise Environmentcited by 4
  29. 2026Nexus: Same Pretraining Loss, Better Downstream Generalization via Common Minimacited by 4
  30. 2026Stochastic KV Routing: Enabling Adaptive Depth-Wise Cache Sharingcited by 4
  31. 2026NemotronLabs VoiceChat: An Open Full-duplex Speech-to-Speech Model with Tool Calling Capabilitiescited by 4
  32. 2025Best-of-∞ -- Asymptotic Performance of Test-Time LLM Ensemblingcited by 3
  33. 2025Learning to Reason: Training LLMs with GPT-OSS or DeepSeek R1 Reasoning Tracescited by 3
  34. 2026Effective Distillation to Hybrid xLSTM Architecturescited by 3
  35. 2026AdaSplash-2: Faster Differentiable Sparse Attentioncited by 3
  36. 2026Marco-MoE: Open Multilingual Mixture-of-Expert Language Models with Efficient Upcyclingcited by 3
  37. 2026Star Elastic: Many-in-One Reasoning LLMs with Efficient Budget Controlcited by 3
  38. 2026Reliable Chain-of-Thought via Prefix Consistencycited by 3

50 most cited of 108 All 108 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025NVIDIA Nemotron Nano V2 VL
  2. 2026Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning
  3. 2024The Mamba in the Llama: Distilling and Accelerating Hybrid Models
  4. 2026Nemotron 3 Nano Omni: Efficient and Open Multimodal Intelligence
  5. 2025Jet-Nemotron: Efficient Language Model with Post Neural Architecture Search
  6. 2026Nautile-370M: Spectral Memory Meets Attention in a Small Reasoning Model
  7. 2026HEBATRON: A Hebrew-Specialized Open-Weight Mixture-of-Experts Language Model
  8. 2026Nemotron-Labs-TwoTower: Diffusion Language Modeling with Pretrained Autoregressive Context