MathArena - Kangaroo 2025 Levels 3-4 — leaderboard
Metric: Accuracy (%). Source: matharena.ai. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Human Expert | 100 |
| 2 | GPT-5.5 (xHigh) | 89.58 |
| 3 | GPT-5.4 (xHigh) | 83.33 |
| 4 | Gemini 3.1 Pro (Preview) | 76.04 |
| 5 | GPT-5.2 (High) | 73.96 |
| 6 | Gemini 3.5 Flash | 72.92 |
| 7 | Qwen 3.5 397B A17B | 68.75 |
| 8 | Gemini 3 Pro (Preview) | 66.67 |
| 9 | GPT-5 Mini (High) | 66.67 |
| 10 | Gemini 3 Flash | 66.67 |
| 11 | GPT-5.1 (High) | 65.62 |
| 12 | Kimi K2.5 (Thinking) | 65.62 |
| 13 | Median Human | 65 |
| 14 | Gemini 2.5 Pro | 64.58 |
| 15 | Claude Sonnet 4.5 (Thinking) | 62.5 |
Interactive version: theaggregate.ai/benchmark?slug=matharena-kangaroo-2025-levels-3-4 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.