NVIDIA analyzes post-training and test-time design for olympiad proof generation
AIMade by AIWritten by AI (Claude) from the source below. Check anything important against the source.How it's made
Analyzed the effect of post-training and test-time reasoning design on natural-language proof generation for olympiad mathematics, and presented an open pipeline that trains two specialized checkpoints on Nemotron 3 Ultra with SFT and RL. It works entirely in natural language, without formal provers, external tools or internet access: a general open model and the two specialized checkpoints run an iterative search that generates, checks and refines candidate proofs, and a separate high-compute stage picks the final submission. The authors claim the system scored 30 out of 42 at IMO 2026, reaching the gold-medal threshold, and released the two checkpoints, the training data, code and submitted answers, and Nemotron-IMO-Bench, a benchmark of 200 olympiad-level problems.
Why it matters
A case of releasing, down to the weights and data, a result that reached the IMO gold-medal range through pure natural-language reasoning without a formal checker — greatly widening the reproducibility of top-level mathematical reasoning.
Source — Primary sources first; secondary reports are labelled “Press”
Paperarxiv.orghttps://arxiv.org/abs/2609.10712Added in Week 38, 2026 · See that week
On the map — ⑥ Reasoning Models line
- Previous stopClaude Fable 5.1
- This recordNVIDIA analyzes post-training and test-time design for olympiad proof generation
- Next stopClaude Sonnet 5.5