MathArena - BRUMO 2025 — leaderboard

Metric: Accuracy (%). Source: matharena.ai. 45 models tracked.

Top models

#ModelScore
1Gemini 3 Flash100
2Step 3.5 Flash100
3Human Expert100
4GLM-599.17
5DeepSeek V3.2 Speciale99.17
6Gemini 3 Pro (Preview)98.33
7Kimi K2.5 (Thinking)98.33
8GPT-5.2 (High)98.33
9Grok 4.1 Fast (Reasoning)97.5
10DeepSeek V3.2 (Thinking)96.67
11Grok 4 Fast (Reasoning)95.83
12O3 (High)95.83
13DeepSeek V3.2 Exp (Thinking)95.83
14Grok 495
15GLM-4.694.17

Interactive version: theaggregate.ai/benchmark?slug=matharena-brumo-2025 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.