MathNet-RAG - Zero-Shot: leaderboard
Metric: on MathNet-RAG, 35 Olympiad problems each paired by experts with a related problem, graded by human experts; accuracy (%) with only the target problem; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 7 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 89.1 |
| 2 | DeepSeek V3.2 Speciale | 84.8 |
| 3 | GPT-5 | 76.8 |
| 4 | Grok 4.1 Fast | 75.4 |
| 5 | Claude Opus 4.5 | 46.8 |
| 6 | OLMo 3 32B (Thinking) | 45.2 |
Interactive version: theaggregate.ai/benchmark?slug=mathnet-rag-zero-shot · How It Works · Data refreshed daily, snapshot 2026-10-07.