AIME: leaderboard
National math exam given to top high-school students.
Metric: Accuracy (%). Source: vals.ai. Saturation forecast: Estimated already saturated. 96 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 98.12 | #58 |
| 2 | GPT-5.2 | 96.88 | #110 |
| 3 | Muse Spark | 96.88 | #64 |
| 4 | Gemini 3 Pro (Preview) | 96.68 | #66 |
| 5 | GPT-5.4 (2026-03-05) | 96.67 | #59 |
| 6 | Grok 4.20 0309 (Reasoning) | 96.46 | #112 (Grok 4.20 0309) |
| 7 | Claude Opus 4.7 | 96.25 | #49 |
| 8 | Gemini 3 Flash (Preview) | 95.62 | #82 |
| 9 | Kimi K2.5 (Thinking) | 95.62 | #144 (Kimi K2.5) |
| 10 | Claude Opus 4.6 (Thinking) | 95.62 | #63 (Claude Opus 4.6) |
| 11 | GPT-5.4 Mini (2026-03-17) | 95.62 | #169 |
| 12 | Claude Opus 4.5 (20251101) (Thinking) | 95.42 | #74 (Claude Opus 4.5 (20251101)) |
| 13 | Qwen 3.6 Plus | 94.58 | #120 |
| 14 | GPT-5 | 93.37 | #98 |
| 15 | GLM-4.7 | 93.33 | #192 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=aime · How It Works · Data refreshed daily, snapshot 2026-10-20.