NEWS · MODELS · #1666
Nemotron-3 specialists fine-tuned to gold level at IOI 2026 and IMO 2026
Teams starting from Nemotron 3 used supervised fine-tuning, reinforcement learning, and an inference loop (GenCorrect) to create specialist checkpoints. Nemotron-3-Ultra-CC scored 535.4/600 in a live IOI 2026 run (unofficial benchmark) and a combined IMO system scored 30/42 on IMO 2026 proofs judged by official graders, exceeding the gold threshold.
KEY POINTS
- Teams starting from Nemotron 3 used supervised fine-tuning, reinforcement learning, and an inference loop (GenCorrect) to create specialist checkpoints.
- Nemotron-3-Ultra-CC scored 535.4/600 in a live IOI 2026 run (unofficial benchmark) and a combined IMO system scored 30/42 on IMO 2026 proofs judged by official graders, exceeding the gold threshold.
- This shows a single foundation model family can be efficiently specialized with standard post-training and search techniques to reach top human-competition performance across distinct reasoning domains.
WHY IT MATTERS
This shows a single foundation model family can be efficiently specialized with standard post-training and search techniques to reach top human-competition performance across distinct reasoning domains.