mlsanityData to 7 Oct 2026

Nemotron-Cascade 2: Post-Training LLMs with Cascade RL and Multi-Domain On-Policy Distillation

arXiv 2603.19220 ↗First posted 19 Mar 2026cs.CL cs.AI cs.LGcited by 53

On the map: LLM agents and tool use → General LLM agent frameworks and benchmarks

also in 1 more
Abstract (arXiv)

We introduce Nemotron-Cascade 2, an open 30B MoE model with 3B activated parameters that delivers best-in-class reasoning and strong agentic capabilities. Despite its compact size, its mathematical and coding reasoning performance approaches that of frontier open models. It is the second open-weight LLM, after DeepSeekV3.2-Speciale-671B-A37B, to achieve Gold Medal-level performance in the 2025 International Mathematical Olympiad (IMO), the International Olympiad in Informatics (IOI), and the ICPC World Finals, demonstrating remarkably high intelligence density with 20x fewer parameters. In contrast to Nemotron-Cascade 1, the key technical advancements are as follows. After SFT on a meticulously curated dataset, we substantially expand Cascade RL to cover a much broader spectrum of reasoning and agentic domains. Furthermore, we introduce multi-domain on-policy distillation from the strongest intermediate teacher models for each domain throughout the Cascade RL process, allowing us to efficiently recover benchmark regressions and sustain strong performance gains along the way. We release the collection of model checkpoint and training data.

Key facts. Cited by 53 papers in the map (data through 7 Oct 2026). Placed in General LLM agent frameworks and benchmarks, one of 16 directions in LLM agents and tool use. First posted on arXiv 19 Mar 2026.

Builds on

0 of 0 references.

No references here yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 53.

53 papers.

Sort:

By area:

  1. 2026Cosmos 3: Omnimodal World Models for Physical AIcited by 137
  2. 2026Uni-OPD: Unifying On-Policy Distillation with a Dual-Perspective Recipecited by 43
  3. 2026Nemotron 3 Ultra: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoningcited by 32
  4. 2026KL for a KL: On-Policy Distillation with Control Variate Baselinecited by 24
  5. 2026Decoupling KL and Trajectories: A Unified Perspective for SFT, DAgger, Offline RL, and OPD in LLM Distillationcited by 16
  6. 2026Counteraction-Aware Multi-Teacher On-Policy Distillation for General Capability Recovery with Domain Preservationcited by 14
  7. 2026Dense Supervision, Sparse Updates: On the Sparsity and Geometry of On-Policy Distillationcited by 10
  8. 2026OneReason Technical Reportcited by 8
  9. 2026PowerOPD: Stabilizing On-Policy Distillation with Bounded Power Transformationcited by 7
  10. 2026UI-MOPD: Multi-Platform On-Policy Distillation for Unified GUI Agentscited by 7
  11. 2026Rethinking Privileged Information in On-Policy Self-Distillationcited by 6
  12. 2026Large Language Model Post-Training: A Unified View of Off-Policy and On-Policy Learningcited by 5
38 more
  1. 2026Towards Understanding On-Policy Distillation through the Lens of Test-Time Scalingcited by 5
  2. 2026Distill What the Student Can See: Fisher-Projected On-Policy Distillation for Vision-Language Modelscited by 4
  3. 2026Verify Before You Distill: Prompt-Level Teacher Gating for On-Policy Distillationcited by 4
  4. 2026The Physics of Multi-Turn Long-Horizon Planning: From Pre-training to Post-training via Single- and Multi-Teacher On-Policy Agentic Distillationcited by 3
  5. 2026Lightning OPD 2.0: Mitigating Style Bias in Cross-Teacher On-Policy Distillation for Large Reasoning Modelscited by 3
  6. 2026Every Coin Has Two Sides: On the Dual Nature of Generalization in On-Policy Distillation of Large Language Modelscited by 3
  7. 2026Same Model, Different Harness: Different Coding-Agent Resultscited by 3
  8. 2026Marco-MoE: Open Multilingual Mixture-of-Expert Language Models with Efficient Upcyclingcited by 2
  9. 2026Your Language Model is Its Own Critic: Reinforcement Learning with Value Estimation from Actor's Internal Statescited by 2
  10. 2026Trust Region Policy Distillationcited by 2
  11. 2026Learn from Whoever Is Right: Answer-Verified Multi-Teacher Distillation for Multi-Domain LLMscited by 2
  12. 2026Eliciting Weak-to-Strong Generalization with On-Policy Reverse Distillationcited by 2
  13. 2026Beyond Token-Local Imitation: Reward-Compatible Temporal Credit Assignment for On-Policy Distillationcited by 2
  14. 2026TMAS: Scaling Test-Time Compute via Multi-Agent Synergycited by 1
  15. 2026EST-PRM: Stress-Testing Process Reward Models Before They Become Load-Bearingcited by 1
  16. 2026Language-Specialized Multi-Teacher On-Policy Distillation for Multilingual LLM-Based ASRcited by 1
  17. 2026Scheduling Mixed RL Rollouts Beyond Prefix Localitycited by 1
  18. 2026D3-MOPD: Dynamic Domain ScheDuling for Efficient Multi-Teacher Distillationcited by 1
  19. 2026Safin-1: Safety from Within through Memory-Native State Evolutioncited by 1
  20. 2026Data-Free On-Policy Distillation: How Far Can We Go Without External Data?cited by 1
  21. 2026From Parameters to Behaviors: A Survey of Model Fusion for Large Language Modelscited by 1
  22. 2026Signal Reshaping for GRPO in Weak-Feedback Agentic Code Repaircited by 0
  23. 2026CARE-RL: Capability-Aware Reinforcement Learning for Mitigating Cross-Domain Conflictscited by 0
  24. 2026ComBench: A Benchmark for Rigorous Proof Reasoning and Constructive Realization in Olympiad-Level Combinatoricscited by 0
  25. 2026ShopX: A Foundation Model for Intent-to-Item Fulfillment in Agentic Shoppingcited by 0
  26. 2026Noisy-Channel Minimum Bayes Risk Decodingcited by 0
  27. 2026Quantization-Aware Healing: A Practical Recipe for Recovering Compressed, 4-Bit LLMscited by 0
  28. 2026Recovering General Capabilities via Uncertainty-Calibrated Multi-Teacher On-Policy Distillationcited by 0
  29. 2026Instella-MoE Technical Reportcited by 0
  30. 2026GANDR: Claim Auditing for Verifiable Legal Answer Generationcited by 0
  31. 2026Lightning Weave: Improving the Accuracy-Efficiency Frontier of Reasoning Models through Capability Compositioncited by 0
  32. 2026ACLArena: Agent Continue Learning in Multi-stage Post-trainingcited by 0
  33. 2026MOPD-Router: Rethinking Teacher Routing in Multi-Teacher On-Policy Distillationcited by 0
  34. 2026Understanding the Synergy between SFT, RLVR, and OPD in LLM Post-Trainingcited by 0
  35. 2026Selecting Diverse SFT Traces Improves Post-RL Generalizationcited by 0
  36. 2026No Pain, More Gain: Iterative Merging for Effective Multi-Teacher On-Policy Distillationcited by 0
  37. 2026PrivacySkills: How Privacy Guidance Shapes Source Selection in LLM Agentscited by 0
  38. 2026ROSS: Relearning from Self-Generated Rollouts through Selective Supervisioncited by 0

50 most cited of 53 All 53 (JSON)

Similar, not cited

Close in wording, no citation either way

  1. 2025Nemotron-Cascade: Scaling Cascaded Reinforcement Learning for General-Purpose Reasoning Models
  2. 2025Nemotron-CrossThink: Scaling Self-Learning beyond Math Reasoning
  3. 2025Llama-Nemotron: Efficient Reasoning Models
  4. 2025Nemotron Elastic: Towards Efficient Many-in-One Reasoning LLMs
  5. 2026Nemotron 3 Super: Open, Efficient Mixture-of-Experts Hybrid Mamba-Transformer Model for Agentic Reasoning
  6. 2025Nemotron-Math: Efficient Long-Context Distillation of Mathematical Reasoning from Multi-Mode Supervision
  7. 2026ReNIO: Reweighting Negative Trajectory Importance for LLM On-Policy Distillation
  8. 2025AceReason-Nemotron: Advancing Math and Code Reasoning through Reinforcement Learning