MathArena - AIME 2026 — leaderboard

Metric: Accuracy (%). Source: matharena.ai. 30 models tracked.

Top models

#ModelScore
1GPT-5.5 (xHigh)100
2Human Expert100
3Claude Opus 4.8 (Max)100
4GPT-5.4 (xHigh)99.17
5Gemini 3.1 Pro (Preview)98.33
6GPT-5.2 (High)98.33
7GLM-596.67
8Gemini 3 Flash96.67
9Step 3.5 Flash96.67
10DeepSeek V4 Pro (Max)96.67
11Claude Opus 4.6 (High)96.67
12GLM-5.195.83
13Kimi K2.5 (Thinking)95.83
14Kimi K2.6 (Thinking)95.83
15DeepSeek V4 Flash (Max)95.83

Interactive version: theaggregate.ai/benchmark?slug=matharena-aime-2026 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.