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
1Claude Fable 5.2max· Epoch AI Benchmarking Hub data95.1%
2GPT-6 Astramax· Epoch AI Benchmarking Hub data93.7%
3GPT-6.1 Solmax· Epoch AI Benchmarking Hub data93.7%
4GPT-6.2 Astramax· Epoch AI Benchmarking Hub data92.6%
5Claude Opus 5.5max· Epoch AI Benchmarking Hub data91.2%
6Gemini 4 Neonxhigh· Epoch AI Benchmarking Hub data91.2%
7GPT-6.1 Terramax· Epoch AI Benchmarking Hub data90.9%
8Gemini 4 Carbonhigh· Epoch AI Benchmarking Hub data90.5%
9Claude Fable 5.1max· Epoch AI Benchmarking Hub data90.2%
10GPT-6 Solmax· Epoch AI Benchmarking Hub data89.8%
11GPT-5.6 Solmax· Epoch AI Benchmarking Hub data89.1%
12Claude Sonnet 5.5max· Epoch AI Benchmarking Hub data88.8%
13Gemini 4 Hydrogenxhigh· Epoch AI Benchmarking Hub data88.1%
14GPT-5.5 Proxhigh· Epoch AI Benchmarking Hub data87.7%
15Claude Fable 5max· Epoch AI Benchmarking Hub data87%
16Gemini 4 Argonhigh· Epoch AI Benchmarking Hub data86%
17GPT-5.6 Terramax· Epoch AI Benchmarking Hub data86%
18Claude Opus 5max· Epoch AI Benchmarking Hub data85.6%
19GPT 5.5xhigh· Epoch AI Benchmarking Hub data85.3%
20GPT-5.4 Proxhigh· Epoch AI Benchmarking Hub data82.5%
21GPT-5.6 Lunamax· Epoch AI Benchmarking Hub data82.1%
22Muse Wonder 1.0max· Epoch AI Benchmarking Hub data82.1%
23Claude Opus 4.8max· Epoch AI Benchmarking Hub data80%
24GPT-6 Lunamax· Epoch AI Benchmarking Hub data78.9%
25GPT 5.4xhigh· Epoch AI Benchmarking Hub data78.6%
26Claude Haiku 5.5max· Epoch AI Benchmarking Hub data75.1%
27Muse Spark 1.3xhigh· Epoch AI Benchmarking Hub data74.4%
28Muse Spark 1.3max· Epoch AI Benchmarking Hub data74%
29GPT-5.2 Proxhigh· Epoch AI Benchmarking Hub data74%
30Kimi K3max· Epoch AI Benchmarking Hub data72.2%
31Gemini 3.7 Flashhigh· Epoch AI Benchmarking Hub data71.6%
32DeepSeek V4.1 Flash 1017max· Epoch AI Benchmarking Hub data71.2%
33Claude Opus 4.7max· Epoch AI Benchmarking Hub data70.2%
34Muse Spark Open 816B A42Bxhigh· Epoch AI Benchmarking Hub data69.8%
35GLM 5.3max· Epoch AI Benchmarking Hub data68.8%
36Gemini 3.8 Flashhigh· Epoch AI Benchmarking Hub data68.4%
37GPT 5.2xhigh· Epoch AI Benchmarking Hub data67.4%
38DeepSeek V4.1 Flashmax· Epoch AI Benchmarking Hub data67.4%
39Claude Opus 4.6max· Epoch AI Benchmarking Hub data66%
40Grok 4.6xhigh· Epoch AI Benchmarking Hub data66%
41Claude Sonnet 5max· Epoch AI Benchmarking Hub data65.6%
42Qwen3.8 Maxxhigh· Epoch AI Benchmarking Hub data65.6%
43DeepSeek V4 Pro 0813max· Epoch AI Benchmarking Hub data64.6%
44Qwen3.7 Max· Epoch AI Benchmarking Hub data64.6%
45Gemini 3.5 Flashhigh· Epoch AI Benchmarking Hub data62.8%
46Gemini 3.1 Pro Preview· Epoch AI Benchmarking Hub data59.6%
47GLM 5.2max· Epoch AI Benchmarking Hub data59.2%
48Gemini 3.6 Flashhigh· Epoch AI Benchmarking Hub data58.9%
49Mistral Large 4high· Epoch AI Benchmarking Hub data58.6%
50DeepSeek V4 Flash 0731max· Epoch AI Benchmarking Hub data57.5%

Showing 1–50 of 125 results.

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