Fine-Tuned Nemotron 3 Models Achieve Gold-Level Scores at IOI and IMO 2026

Fine-Tuned Nemotron 3 Models Achieve Gold-Level Scores at IOI and IMO 2026 NVIDIA applied supervised fine-tuning, reinforcement learning, and feedback-driven inference loops to Nemotron 3 foundation models, earning gold-medal-level scores on both the IOI 2026 and IMO 2026 benchmarks. Nemotron-3-Ultra-CC scored 535.4/600 on IOI 2026, exceeding the gold threshold of 361.12 and the top human score of 498.27 in an unofficial prospective run. A multi-checkpoint generate-verify-refine system scored 30/42 on IMO 2026, evaluated by official graders, surpassing the 29-point gold threshold. Specialization relied on a reusable four-part framework: strong base models, curated domain traces, SFT/RL post-training, and test-time evaluation loops. IOI specialization paired curated problems with the GenCorrect refinement strategy across Nano (30B/3B active) and Ultra (550B/55B active) architectures. The IMO system operated entirely in natural language without formal provers or external tools, combining complementary SFT and RL models. Reusable Post-Training Framework The adaptation framework avoids training new foundation models from scratch by relying on a four-part recipe: starting from a Nemotron base model, curating domain problems and reasoning traces, applying post-training methods like SFT and RL, and pairing the specialist model with an iterative generation-and-verification loop during inference. Competitive Programming Specialization on IOI For competitive programming, training utilized 22,000 problems alongside synthetic reasoning traces. Nemotron-3-Nano-CC (30B total, 3B active parameters) received SFT and RL, while Nemotron-3-Ultra-CC (550B total, 55B active parameters) received SFT. Paired with GenCorrect—an iterative generate-evaluate-refine strategy—Ultra-CC achieved 535.4 out of 600 points on IOI 2026 under the same time and submission constraints as human competitors, though running as an unofficial, unsupervised benchmark. Natural-Language Mathematics on IMO For the IMO, post-training used Nemotron 3 Ultra to produce complementary SFT and RL checkpoints. The SFT dataset comprised 414,890 filtered examples over 15,818 problems covering generation, refinement, verification, and meta-verification, while RL focused on 9,597 problems near the capability boundary. The system constructed, critiqued, and refined arguments purely in natural language without formal theorem provers, scoring 30 out of 42 with full credit on four of six problems as graded by official IMO graders. Interaction of Specialization and Test-Time Compute The results demonstrate that test-time search benefits from targeted model specialization rather than brute-force sampling alone. Stronger domain-specialized models produce superior candidates, critics, and revisions for inference search loops like GenCorrect and generate-verify-refine pipelines. Open-Source Releases Related assets have been made publicly available. Releases include the Nemotron-3-Ultra-CC model, the Nemotron Labs IMO 2026 checkpoints and datasets, the Nemotron-IMO-Bench dataset of 200 problems, technical papers, and inference scripts in the NeMo-Skills repository.

Nemotron系列模型在IOI与IMO竞赛中达到金牌水平 基于Nemotron 3基础模型，研究团队通过监督微调（SFT）、强化学习（RL）及反馈驱动的推理循环，构建出在IOI 2026与IMO 2026均达到金牌表现的专业系统。 Nemotron-3-Ultra-CC在IOI 2026中取得535.4/600分，超过金牌线361.12分及人类最高分498.27分（非官方基准测试）。 Nemotron 3 Ultra在IMO 2026中获得30/42分，官方评卷评分超过29分金牌线，全程使用自然语言且未调用形式化求解器。 两项成果均依赖四步可复用流程：强基础模型、领域数据构建、后训练（SFT/RL）以及结合评估优化的推理循环。 相关模型、训练数据集、推理流水线及论文已在Hugging Face和NeMo-Skills等平台开源。 专业化微调与IOI竞赛表现 针对编程竞赛，团队筛选了2.2万个题目并生成合成推理轨迹。模型规模包括总参数300亿（活跃30亿）的Nano版本和总参数5500亿（活跃550亿）的Ultra版本。Nano版本在SFT与RL后结合迭代修正策略GenCorrect，在IOI 2025测试中达到468分；而Ultra版本仅凭一轮SFT与GenCorrect便在IOI 2026取得535.4分的高分。 自然语言数学证明与IMO竞赛 在IMO项目中，团队训练了基于41.4万条多阶段（生成、修正、验证等）数据的SFT检查点，以及基于近1万道能力边界题目的RL检查点。最终系统完全以自然语言运行，不依赖外部工具或网络，通过生成、评分、批评和提炼候选证明，在6道题中获得4道满分，总分达到30分。 模型微调与测试时计算的协同 结果表明金牌成绩并非仅靠单独微调或盲目暴力采样，而是源于模型、数据与推理循环的协同设计。微调为推理系统提供了更高质量的候选答案与评估基准，而IOI的GenCorrect多轮反馈与IMO的生成-验证-提炼系统则将这些专业能力充分放大。

The International Olympiad in Informatics (IOI) and the International Mathematical Olympiad (IMO) test different skills. IOI requires algorithms and code that pass hidden tests under strict time and submission limits. IMO demands rigorous natural-language proofs. Success at either competition is difficult. Success at both points to something broader. Our recent results show that Nemotron is a strong, adaptable foundation for building world-class specialist models. Starting from Nemotron 3, our teams used supervised fine-tuning (SFT), reinforcement learning (RL), and feedback-driven inference to create systems that reached gold-medal level at both IMO 2026 and IOI 2026 . Competition Nemotron specialization Result IOI 2026 Nemotron-3-Ultra-CC with SFT and GenCorrect 535.4/600, above the 361.12 gold threshold and the top human score of 498.27 IMO 2026 Nemotron 3 Ultra general, SFT, and RL checkpoints in a generate-verify-refine system 30/42, above the official gold threshold of 29 The IOI result came from a live, prospective run under the same time, internet-access, and submission constraints as human contestants. It was an unofficial, unsupervised benchmark and was not included in the official IOI ranking. The IMO system’s submitted proofs were graded by official IMO graders. A reusable specialization recipe "Easy to fine-tune" should mean more than making a checkpoint trainable. It should mean that a capable foundation model can be adapted to a demanding domain with a clear, reusable recipe. Across the two projects, that recipe had four parts: Start with a strong Nemotron base model. Curate domain-specific problems and high-quality reasoning traces. Apply standard post-training methods such as SFT and, where useful, RL. Pair the specialist model with an inference loop that generates, evaluates, and improves candidate answers. The training and inference runs were substantial, but the underlying approach is familiar and reproducible. We did not need to build a new foundation model for every challenge. We specialized Nemotron for the task. From general coding ability to IOI gold For competitive programming, we curated 22,000 problems and generated synthetic reasoning traces to train two specialists. Nemotron-3-Nano-CC, with 30 billion total parameters and 3 billion active parameters, received both SFT and RL. Nemotron-3-Ultra-CC, with 550 billion total parameters and 55 billion active parameters, received SFT. The progression on IOI 2025 makes the value of specialization easy to see. Nano improved from 130 points before post-training to 280 after SFT and 291 after RL. With GenCorrect, our iterative generate-evaluate-refine strategy, it reached 468 points and crossed the gold threshold of 438.3. Ultra-CC reached 502 points with the same test-time strategy. These experiments also showed that adaptation does not have to look the same at every scale. SFT produced most of Nano's gain, with RL adding a smaller but consistent improvement. For the stronger Ultra model, one SFT epoch was enough to outperform the fully post-trained Nano model across IOI, ICPC, and LiveCodeBench Pro. That finding guided the competition-specific Ultra-CC system used for IOI 2026, which scored 535.4 out of 600. Teaching Nemotron to prove, check, and revise The IMO project applied the same idea to olympiad mathematics. Starting from Nemotron 3 Ultra, we trained one specialist with SFT and another with RL. The SFT corpus contained 414,890 quality-filtered examples across 15,818 unique proof problems. It did more than teach final answers. The data covered proof generation, refinement, verification, and meta-verification, so the model learned to construct arguments, identify gaps, respond to critiques, and judge whether a proof was complete. The RL model was trained on 9,597 proof problems selected near the model's capability frontier. Both post-trained checkpoints outperformed the general-availability model in the development experiments. The SFT checkpoint was strongest in the first search round, while the RL checkpoint achieved the best overall single-checkpoint result. Their strengths were complementary, so the final system used both specialists alongside the general model. For each IMO problem, the models generated candidate proofs, scored them, produced critiques, and refined the most promising attempts. A separate high-compute stage selected the final submission. The entire system worked in natural language, with no formal prover, external tools, or internet access. It scored 30 out of 42 points, including full credit on four of the six problems, and exceeded the official gold-medal threshold. Fine-tuning and test-time compute work together Our earlier IOI 2025 Hugging Face post showed how test-time compute can push open-weight models to gold-level performance. The new results add an important piece: better specialization gives the inference system better candidates, better critics, and better refinements. At IOI, GenCorrect turned the gains from fine-tuning into larger improvements over multiple feedback rounds. At IMO, using complementary SFT and RL checkpoints was more valuable than simply drawing more samples from one checkpoint. In both cases, the best outcome came from combining a capable specialist with a system that could search, verify, and improve. This distinction matters. The medals were not produced by fine-tuning alone, and they were not produced by brute-force sampling alone. They came from co-designing the model, the data, and the inference loop. Open models, data, and recipes on Hugging Face We want these results to be useful beyond the competitions. The Nemotron Labs IMO 2026 collection brings together the SFT and RL checkpoints, both training datasets, and Nemotron-IMO-Bench, a new benchmark of 200 olympiad-level problems. The IMO paper describes the training approach and generate-verify-refine system, while the NeMo-Skills repository includes the IMO inference pipeline, prompts, submitted proofs, and a reproducible quickstart. For competitive programming, the Nemotron-3-Ultra-CC model is available on Hugging Face, and the IOI paper provides the training recipe and the GenCorrect methodology. The IOI evaluation and inference pipeline are also available in NeMo-Skills . Together, IMO and IOI provide unusually demanding evidence for a simple idea: Nemotron can be fine-tuned into world-class domain specialists, then composed with transparent inference workflows to solve problems at the frontier of human competition. We are excited to see what the Hugging Face community builds next.