MathNet-RAG - Embedding Retrieval: leaderboard
Metric: on MathNet-RAG, 35 Olympiad problems each paired by experts with a related problem, graded by human experts; accuracy (%) with one related problem and its official solution retrieved by gemini-embedding-001; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 7 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 92.9 |
| 2 | DeepSeek V3.2 Speciale | 89.5 |
| 3 | Grok 4.1 Fast | 83.8 |
| 4 | GPT-5 | 75.2 |
| 5 | Claude Opus 4.5 | 55.5 |
| 6 | OLMo 3 32B (Thinking) | 54.6 |
Interactive version: theaggregate.ai/benchmark?slug=mathnet-rag-embedding-retrieval · How It Works · Data refreshed daily, snapshot 2026-10-07.