Paper 2603.19220
Nemotron-Cascade 2: Post-Training LLMs with Cascade RL and Multi-Domain On-Policy Distillation
- Published
- Mar 2026
- Research lab
- NVIDIA
- Citations
- 20
- GitHub
- Not linked
01 In brief
Summary
Nemotron-Cascade 2 is an open 30B Mixture-of-Experts (MoE) model with 3B activated parameters, achieving best-in-class reasoning and agentic capabilities.
It is the second open-weight LLM to achieve Gold Medal-level performance in the 2025 IMO, IOI, and ICPC World Finals, with 20x fewer parameters than DeepSeek-V3.2-Speciale-671B-A37B.
The model builds on Nemotron-Cascade 1 by expanding Cascade RL to cover more reasoning and agentic domains and introducing multi-domain on-policy distillation (MOPD) from the strongest intermediate teacher models per domain, which recovers benchmark regressions and sustains performance gains.
The post-training pipeline includes SFT, IF-RL, multi-domain RL, MOPD, RLHF, long-context RL, code RL, and SWE RL.
The model outperforms Qwen3.5-35B-A3B and Nemotron-3-Super-120B-A12B on math, code reasoning, alignment, and instruction following, but underperforms on knowledge-intensive and agentic benchmarks.
The model and training data are open-sourced.
02 From the paper
Abstract
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.