MMODELYST
CAP

Math

League table — best at math · 30
mean percentile across this capability's benchmarks
#ModelLeagueCoverage$/1Mtok/s
1GPT-5.2100.01/4$4.8181
2AceMath-72B-Instruct100.01/4
3AceMath-7B-Instruct99.91/4
4AceInstruct-72B99.91/4
5Qwen2.5-32B-Instruct99.81/4
6Qwen2.5-Math-72B-Instruct99.71/4
7Awqward2.5-32B-Instruct99.71/4
8Linkbricks-Horizon-AI-Avengers-V6-32B99.61/4
9Linkbricks-Horizon-AI-Avengers-V3-32B99.51/4
10GPT-5 Codex99.51/4$3.44175
11lambda-qwen2.5-32b-dpo-test99.51/4
12Linkbricks-Horizon-AI-Avengers-V1-32B99.41/4
13Qwen2.5-72B-Instruct-abliterated99.31/4
14WomboCombo-R1-Coder-14B-Preview99.31/4
15qwen2.5-test-32b-it99.11/4
16Qwen2.5-32B-Instruct-abliterated-v299.11/4
17Gemini 3 Flash Preview99.01/4$1.13202
18huihui-ai-abliterated-Qwen2.5-32B-Inst-BaseMerge-TIES99.01/4
19calme-2.1-qwen2.5-72b98.91/4
20calme-2.2-qwen2.5-72b98.91/4
21zetasepic-abliteratedV2-Qwen2.5-32B-Inst-BaseMerge-TIES98.81/4
22Qwen2.5-72B-2x-Instruct-TIES-v1.098.71/4
23Lion-Lamarck-v.1.1.098.61/4
24Gauss-Opus-14B-R99998.61/4
25DeepSeek V3.2 Speciale98.51/4
26DeepSeek-R1-Distill-Qwen-14B98.51/4
27Sombrero-Opus-14B-Sm198.41/4
28Linkbricks-Horizon-AI-Avengers-V2-32B98.41/4
29Viper-Coder-v1.6-r99998.31/4
30o-distil-qwen98.21/4

League = a model's mean percentile across the 4 benchmarksin this capability (best score per benchmark; partial coverage shown). Price & speed via Artificial Analysis.

Benchmarks in this capability (4)
AIME209 scores
GSM8K10 scores
MATH192 scores
MATH Lvl 51480 scores