MathArena - APEX 2025: leaderboard

Metric: Accuracy (%). Source: matharena.ai. 48 models tracked.

Top models

#ModelScore
1Claude Opus 4.8 (Max)81.25
2GPT-5.5 (xHigh)80.21
3Hy4 preview74.2
4GPT-5.4 Pro (xHigh)69.79
5Gemini 3.1 Pro (Preview)60.94
6GPT-5.4 (xHigh)54.17
7Qwen 3.7 Max44.5
8Claude Opus 4.7 (xHigh)40.62
9DeepSeek V4 Pro38.3
10Nex N2 Pro36.5
11Claude Opus 4.6 (Max)34.5
12Claude Opus 4.6 (High)34.45
13Gemini 3.5 Flash32.29
14DeepSeek V4 Pro (Max)28.12
15DeepSeek V4 Flash (Max)27.08

Interactive version: theaggregate.ai/benchmark?slug=matharena-apex-2025 · How It Works · Data refreshed daily, snapshot 2026-09-05.