Skip to content

Benchmark

Frontiermath Tiers 1 3 V2

Math benchmarks measure how well models solve maths problems. Scores are percentages.

Models measured
93
Scores recorded
125
Sources
1
IQ weight
1.66

Score distribution

47.3%
Mean
45.3%
Median
28.4%
Std dev
9.5%
10th percentile
88.1%
90th percentile
0%
Min
0%95.1%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.66 for chat models, measured over 93 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
51Grok 4.5high· Epoch AI Benchmarking Hub data57.2%
52Kimi K2.6· Epoch AI Benchmarking Hub data57.2%
53GLM 5.3 Flashmax· Epoch AI Benchmarking Hub data55.8%
54GPT-5 Prohigh· Epoch AI Benchmarking Hub data55.8%
55GPT 5high· Epoch AI Benchmarking Hub data55.4%
56GLM 5.2low· Epoch AI Benchmarking Hub data54.7%
57Kimi K2.7 Code· Epoch AI Benchmarking Hub data54%
58Grok 4.7xhigh· Epoch AI Benchmarking Hub data53%
59Gemini 3 Flash· Epoch AI Benchmarking Hub data51.2%
60GPT-5.4 minixhigh· Epoch AI Benchmarking Hub data51.2%
61GPT-5 minihigh· Epoch AI Benchmarking Hub data46.7%
62Inkling Smallxhigh· Epoch AI Benchmarking Hub data46.3%
63DeepSeek V4 Promax· Epoch AI Benchmarking Hub data45.3%
64GPT 5.4 Nanohigh· Epoch AI Benchmarking Hub data44.9%
65Grok 4.20reasoning, 0309· Epoch AI Benchmarking Hub data44.9%
66Grok 4.3high· Epoch AI Benchmarking Hub data42.8%
67GLM 5.2none· Epoch AI Benchmarking Hub data42.5%
68GPT-5.6 Lunalow· Epoch AI Benchmarking Hub data41.4%
69GPT-5.6 Lunanone· Epoch AI Benchmarking Hub data39.6%
70Qwen3.6 Plus· Epoch AI Benchmarking Hub data38.2%
71GPT 5low· Epoch AI Benchmarking Hub data37.2%
72GLM 5.1· Epoch AI Benchmarking Hub data36.8%
73o4 Minihigh· Epoch AI Benchmarking Hub data36.1%
74Qwen3.6 27B· Epoch AI Benchmarking Hub data35.1%
75Claude Opus 4.532k· Epoch AI Benchmarking Hub data34.4%
76Qwen3.7 Plusnone· Epoch AI Benchmarking Hub data34.4%
77Qwen3.6 27Bnone· Epoch AI Benchmarking Hub data34%
78Inkling FP4xhigh· Epoch AI Benchmarking Hub data33.3%
79OpenAI o3high· Epoch AI Benchmarking Hub data33.3%
80Qwen3.6 Plusnone· Epoch AI Benchmarking Hub data32.3%
81Qwen3.5 397B A17Bnone· Epoch AI Benchmarking Hub data31.2%
82OpenAI o3medium· Epoch AI Benchmarking Hub data29.8%
83Qwen3.5 397B A17B· Epoch AI Benchmarking Hub data29.5%
84o4 Minimedium· Epoch AI Benchmarking Hub data28.8%
85Gemini 3.1 Flash Litehigh· Epoch AI Benchmarking Hub data27.7%
86GPT-5.5 Instant· Epoch AI Benchmarking Hub data26.3%
87Gemini 3.5 Flash Litehigh· Epoch AI Benchmarking Hub data26%
88GLM 5.1none· Epoch AI Benchmarking Hub data24.9%
89Gemini 2.5 Pro· Epoch AI Benchmarking Hub data24.6%
90GPT-5.4 minilow· Epoch AI Benchmarking Hub data24.6%
91Claude Sonnet 4.532k· Epoch AI Benchmarking Hub data23.9%
92Gemini 3.1 Flash Litelow· Epoch AI Benchmarking Hub data22.5%
93Qwen3.6 Flashnone· Epoch AI Benchmarking Hub data22.5%
94Gemini 3.1 Flash Liteminimal· Epoch AI Benchmarking Hub data21.4%
95GPT 5.4 Nanolow· Epoch AI Benchmarking Hub data20.4%
96Qwen3.6 35B A3Bnone· Epoch AI Benchmarking Hub data20.4%
97GPT-5 nanohigh· Epoch AI Benchmarking Hub data20%
98OpenAI o3low· Epoch AI Benchmarking Hub data19.3%
99Qwen3.7 Flash· Epoch AI Benchmarking Hub data19.3%
100Qwen3 Max· Epoch AI Benchmarking Hub data18.9%

Showing 51–100 of 125 results.

Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence)