FrontierMath - Tier 4 — leaderboard
Metric: Accuracy (%, 48 problems). Source: epoch.ai. 72 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 Pro (xHigh) | 39.6 |
| 2 | GPT-5.4 Pro (xHigh) | 37.5 |
| 3 | GPT-5.5 (xHigh) | 35.4 |
| 4 | GPT-5.2 Pro | 31.3 |
| 5 | Claude Opus 4.8 (Max) | 31.25 |
| 6 | GPT-5.4 (xHigh) | 27.1 |
| 7 | Claude Opus 4.7 (xHigh) | 22.92 |
| 8 | Claude Opus 4.6 (Max) | 22.9 |
| 9 | Claude Opus 4.6 (Thinking 32K) | 20.83 |
| 10 | GPT-5.2 (High) | 18.8 |
| 11 | GPT-5.2 (xHigh) | 18.8 |
| 12 | Gemini 3 Pro (Preview) | 18.75 |
| 13 | Gemini 3.1 Pro (Preview) | 16.7 |
| 14 | GPT-5.2 (Medium) | 16.7 |
| 15 | Muse Spark | 14.6 |
Interactive version: theaggregate.ai/benchmark?slug=frontiermath-tier-4 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.