AGI-Eval Community Base - Mathematical Reasoning: leaderboard

AGI-Eval Community private-set Base Language Models evaluation (Mathematical Reasoning) measuring accuracy (%) across bilingual capability dimensions.

Metric: Accuracy (%). Source: agi-eval.org. Saturation forecast: Estimated already saturated. 25 models tracked. Updated 22 Nov 2026.

Top models

#ModelScoreOverall rank
1ERNIE-4.5-300B-A47B-Base72.53
2Kimi K2 Base70.49#939
3DeepSeek-V3.2-Exp-Base62.42
4DeepSeek-V3.1-Base61.31
5GLM-4.5-Base60.35
6Qwen3-30B-A3B-Base59.83#1475
7Llama 4 Maverick Base59.29
8DeepSeek V3 Base59.25
9Qwen2.5-72B-Base58.41
10Qwen3-14B-Base57.12#1473
11Qwen 2.5 32B (Base)57.08
12Qwen3-8B-Base52.31#1754
13Qwen2.5-14B-Base50.18
14Qwen3-4B-Base48.38#2439
15Qwen 2.5 7B (Base)45.89

No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=agi-eval-community-base-mathematical-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-22.