MathArena - SMT 2025 — leaderboard

Metric: Accuracy (%). Source: matharena.ai. 43 models tracked.

Top models

#ModelScore
1Gemini 3 Pro (Preview)93.4
2Gemini 3 Flash92.92
3GPT-5 (High)91.98
4GPT-5.2 (High)91.98
5Step 3.5 Flash91.51
6GLM-591.04
7Kimi K2 (Thinking)91.04
8GPT-5.1 (High)91.04
9GLM-4.690.57
10Kimi K2.5 (Thinking)90.57
11GPT-5 Mini (High)89.15
12DeepSeek V3.2 Speciale89.15
13O4 Mini (High)88.68
14GPT-OSS-120B (High)87.74
15DeepSeek V3.2 (Thinking)87.74

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