FrontierMath Tier 4

The hardest FrontierMath tier: problems that take specialists days. Run by Epoch AI.

CategoryMath
MetricAccuracy
SourceEpoch AI (evaluator)
LicenseCC BY 4.0
Benchmark pageepoch.ai
StatusScored
Weight1.2 within Math
FitDiscrimination 3.69, noise 0.49
LineageShares weight with frontiermath_tier_4

Results

#ModelProviderResultDate
1 Claude Fable 5.2 Max Anthropic 100.0% 2026-11-06
2 GPT-6.1 Sol Max OpenAI 100.0% 2026-09-29
3 GPT-6 Astra Max OpenAI 97.6% 2026-08-30
4 GPT-6 Astra High OpenAI 97.6% 2026-08-30
5 GPT-6 Astra Extra High OpenAI 97.6% 2026-08-30
6 GPT-6 Astra Medium OpenAI 97.6% 2026-08-30
7 Gemini 4 Carbon High Google 95.1% 2026-11-02
8 Claude Opus 5.5 Max Anthropic 95.0% 2026-09-22
9 GPT-6.1 Terra Max OpenAI 92.7% 2026-10-26
10 Claude Fable 5 Max Anthropic 90.2% 2026-06-09
11 GPT-6 Sol Max OpenAI 90.0% 2026-09-22
12 Claude Fable 5.1 Max Anthropic 87.8% 2026-09-01
13 GPT-6 Astra Low OpenAI 87.8% 2026-08-30
14 GPT-5.6 Sol Max OpenAI 82.9% 2026-07-09
15 GPT-6 Astra Non-reasoning OpenAI 82.9% 2026-08-30
16 Claude Sonnet 5.5 Max Anthropic 80.5% 2026-09-29
17 GPT-5.6 Sol Pro Max OpenAI 80.5% 2026-07-09
18 Gemini 4 Argon High Google 78.1% 2026-10-13
19 GPT-5.5 Pro Extra High OpenAI 78.1% 2026-06-12
20 AI Co-Mathematician Google 75.6% 2026-06-12
21 Claude Opus 5 Max Anthropic 73.2% 2026-07-24
22 GPT-5.5 Extra High OpenAI 72.5% 2026-06-11
23 GPT-5.6 Terra Max OpenAI 70.7% 2026-07-09
24 Muse Wonder Max Meta 70.7% 2026-10-12
25 Kimi K3.1 Pro Max Moonshot AI 65.9% 2026-11-06
26 GPT-5.6 Luna Max OpenAI 61.0% 2026-07-09
27 GPT-5.4 Pro Extra High OpenAI 58.5% 2026-06-13
28 GPT-6 Luna Max OpenAI 56.1% 2026-09-22
29 Claude Opus 4.8 Max Anthropic 56.1% 2026-06-10
30 GPT-5.4 Extra High OpenAI 49.0% 2026-06-11
31 DeepSeek V4.1 Flash 1017 Max DeepSeek 46.3% 2026-10-19
32 Claude Haiku 5.5 Max Anthropic 46.3% 2026-10-07
33 Qwen3.8 Max Extra High Alibaba 46.3% 2026-08-04
34 Muse Spark 1.3 Max Meta 46.3% 2026-09-18
35 GPT-5.2 Pro Extra High OpenAI 46.0% 2026-06-13
36 Muse Spark 1.3 Extra High Meta 41.5% 2026-09-16
37 Kimi K3 Max Moonshot AI 39.0% 2026-07-17
38 Gemini 3.7 Flash High Google 36.6% 2026-08-14
39 Qwen3.7 Max Alibaba 34.2% 2026-06-13
40 Qwen3.8 Max 0902 Extra High Alibaba 34.2% 2026-09-02
41 Muse Spark Open Extra High Meta 31.7% 2026-10-15
42 Grok 4.6 Extra High xAI 31.7% 2026-08-14
43 Claude Opus 4.7 Max Anthropic 31.7% 2026-06-10
44 GPT-5.2 Extra High OpenAI 31.7% 2026-06-11
45 GLM-5.3 Max Zhipu AI 29.3% 2026-08-25
46 Claude Sonnet 5 Max Anthropic 29.3% 2026-06-30
47 GLM-5.2 Max Zhipu AI 29.3% 2026-06-19
48 Gemini 3.1 Pro Preview Google 26.8% 2026-06-11
49 Mistral Large 4 Preview High Mistral 26.8% 2026-10-12
50 DeepSeek V4.1 Flash Max DeepSeek 26.8% 2026-10-08
51 DeepSeek V4 Pro 0813 Max DeepSeek 26.8% 2026-08-19
52 Claude Opus 4.6 Max Anthropic 26.8% 2026-06-11
53 Gemini 3.5 Flash High Google 26.8% 2026-06-10
54 Kimi K2.6 Moonshot AI 25.6% 2026-06-10
55 DeepSeek V4 Flash 0731 Max DeepSeek 24.4% 2026-08-02
56 Grok 4.5 High xAI 24.4% 2026-07-09
57 Gemini 3.8 Flash High Google 22.0% 2026-09-02
58 Gemini 3.6 Flash High Google 22.0% 2026-08-02
59 GPT-5 High OpenAI 22.0% 2026-06-11
60 GPT-5 Pro High OpenAI 19.5% 2026-06-12
61 Grok 4.20 (Reasoning) xAI 17.1% 2026-07-13
62 Gemini 3 Flash Preview Google 17.1% 2026-06-11
63 Grok 4.7 Extra High xAI 17.1% 2026-09-22
64 GLM-5.3-Flash Max Zhipu AI 17.1% 2026-08-27
65 Inkling Small Extra High Thinking Machines 17.1% 2026-08-14
66 Grok 4.3 High xAI 14.6% 2026-06-17
67 Kimi K2.7 Code Moonshot AI 12.2% 2026-06-13
68 GPT-5.4 nano High OpenAI 12.2% 2026-06-12
69 GPT-5 Mini High OpenAI 12.2% 2026-06-12
70 GPT-5.4 mini Extra High OpenAI 9.8% 2026-06-12
71 Inkling Extra High Thinking Machines 4.9% 2026-08-06
72 Claude Opus 4.5 Thinking Anthropic 4.9% 2026-06-11
73 o4-mini High OpenAI 4.9% 2026-06-11
74 GPT-5.5 Instant OpenAI 2.4% 2026-08-02
75 GPT-5 Nano High OpenAI 2.4% 2026-06-12
76 Claude Sonnet 4.5 Thinking Anthropic 2.4% 2026-06-11
77 Claude Opus 4.1 Thinking Anthropic 2.4% 2026-06-11
78 DeepSeek V4 Pro Max DeepSeek 2.4% 2026-06-17
79 Gemini 2.5 Pro Google 0.0% 2026-06-11
80 Gemini 3.5 Flash Lite High Google 0.0% 2026-08-02
81 o3-mini High OpenAI 0.0% 2026-06-11