SignalTech/AI
One Model Family, Two Gold-Level Results: Fine-Tuning Nemotron for IOI and IMO
High-priority alert based on source significance and keyword relevance.
Context
What happened: NVIDIA says Nemotron 3 models, adapted with supervised fine-tuning, reinforcement learning, and generate-verify-refine inference, reached gold-medal level at IOI 2026 and IMO 2026.
Key numbers: IOI 2026: 535.4/600, above the 361.12 gold threshold and top human score of 498.27.
Why it matters: The teams say specialization plus an inference loop, not fine-tuning or sampling alone, produced the results. They released checkpoints, datasets, and a 200-problem Nemotron-IMO-Bench. The IOI run was unofficial and excluded from official ranking.
What changed
NVIDIA says Nemotron 3 models, adapted with supervised fine-tuning, reinforcement learning, and generate-verify-refine inference, reached gold-medal level at IOI 2026 and IMO 2026.
Why it matters
The teams say specialization plus an inference loop, not fine-tuning or sampling alone, produced the results. They released checkpoints, datasets, and a 200-problem Nemotron-IMO-Bench. The IOI run was unofficial and excluded from official ranking.