AoPS Strategy Coverage: leaderboard
Strategy diversity in competition mathematics (Yang et al., 2026): asked to solve 80 AMC 10/12 and AIME problems in as many distinct ways as it can, a model is credited for each of 217 human solution-strategy families from Art of Problem Solving that one of its valid solutions matches, with AI coding and human adjudication. Percent of the reference strategies recovered; higher is better.
Metric: AoPS-reference strategies recovered (% of the 217 reference strategy families for 80 AMC 10/12 and AIME problems; multiple-strategy prompt, one query per problem; dual-AI coding with human adjudication). Source: arxiv.org. Saturation forecast: Around February 2027. 4 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 69.12 | #91 |
| 2 | GPT-5.4 | 59.91 | #121 |
| 3 | DeepSeek V3.2 (Thinking) | 58.99 | #403 (DeepSeek V3.2) |
| 4 | Claude Opus 4.6 | 46.08 | #100 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=aops-strategy-coverage · How It Works · Data refreshed daily, snapshot 2026-11-12.