Skip to main content

AI Rankings

FrontierMath (Tiers 1-3) Leaderboard: AI Models Ranked by Score

This page ranks AI models by their FrontierMath (Tiers 1-3) score, using Epoch AI's open benchmark data.

Data updated

What FrontierMath (Tiers 1-3) measures

Original, very hard maths problems written by mathematicians. Each has an exact answer and the problems are kept private so models cannot have memorised them. This is the Tiers 1 to 3 set, version 2.

Scores are percentages and higher is better. Results come from Epoch AI's open benchmark data; we do not run the test ourselves.

Top 20 at a glance

Tap a model to see its full page: scores, price and how it compares.
Top 20 models on FrontierMath (Tiers 1-3) scoreRanked list of 20 items. 1. GPT-6 Astra, 93.7%; 1. GPT-6.1 Sol, 93.7%; 3. Claude Opus 5.5, 91.2%.
  1. GPT-6 Astra93.7%Released 3 Sep 2026
  2. GPT-6.1 Sol93.7%Released 29 Sep 2026
  3. Claude Opus 5.591.2%Released 22 Sep 2026
  4. Claude Fable 5.190.2%Released 1 Sep 2026
  5. GPT-6 Sol89.8%Released 22 Sep 2026
  6. GPT-5.6 Sol89.1%Released 9 Jul 2026
  7. Claude Sonnet 5.588.8%Released 28 Sep 2026
  8. GPT-5.5 Pro87.7%Released 23 Apr 2026
  9. Claude Fable 587.0%Released 9 Jun 2026
  10. Gemini 4 Argon86.0%Released 30 Sep 2026
  11. GPT-5.6 Terra86.0%Released 9 Jul 2026
  12. Claude Opus 585.6%Released 24 Jul 2026
  13. GPT-5.585.3%Released 23 Apr 2026
  14. GPT-5.4 Pro82.5%Released 5 Mar 2026
  15. GPT-5.6 Luna82.1%Released 9 Jul 2026
  16. Muse Wonder82.1%Released 12 Oct 2026
  17. Claude Opus 4.880.0%Released 28 May 2026
  18. GPT-6 Luna79.0%Released 22 Sep 2026
  19. GPT-5.478.6%Released 5 Mar 2026
  20. Claude Haiku 5.575.1%Released 7 Oct 2026

Full ranking

All 87 models.Download the full table (CSV)Data credit: Epoch AI, LiteLLM model price list, used under open licences (details at the foot of the page).
FrontierMath (Tiers 1-3) Leaderboard: AI Models Ranked by Score: 87 of 87 models
Rank and modelFrontierMath (Tiers 1-3) scoreEpoch indexPrice per 1M tokens (in / out)Released
1GPT-6 AstraOpenAI
93.7%168.8$10.00 / $50.002026-09-03
1GPT-6.1 SolOpenAI
93.7%168.4$2.00 / $10.002026-09-29
3Claude Opus 5.5Anthropic
91.2%168.8$4.00 / $20.002026-09-22
4Claude Fable 5.1Anthropic
90.2%166.5$10.00 / $50.002026-09-01
5GPT-6 SolOpenAI
89.8%163.7$2.00 / $10.002026-09-22
6GPT-5.6 SolOpenAI
89.1%162.8$4.00 / $20.002026-07-09
7Claude Sonnet 5.5Anthropic
88.8%166.2$2.00 / $10.002026-09-28
8GPT-5.5 ProOpenAI
87.7%162.9$30.00 / $1802026-04-23
9Claude Fable 5Anthropic
87.0%163.4$10.00 / $50.002026-06-09
10Gemini 4 ArgonGoogle
86.0%166.4$2.00 / $10.002026-09-30
10GPT-5.6 TerraOpenAI
86.0%160.1$2.00 / $12.002026-07-09
12Claude Opus 5Anthropic
85.6%163.9$5.00 / $25.002026-07-24
13GPT-5.5OpenAI
85.3%159.7$5.00 / $30.002026-04-23
14GPT-5.4 ProOpenAI
82.5%159.6$30.00 / $1802026-03-05
15GPT-5.6 LunaOpenAI
82.1%156.5$0.20 / $1.202026-07-09
15Muse WonderMeta
82.1%161.8Not measured2026-10-12
17Claude Opus 4.8Anthropic
80.0%159.0$5.00 / $25.002026-05-28
18GPT-6 LunaOpenAI
79.0%156.8$0.10 / $0.502026-09-22
19GPT-5.4OpenAI
78.6%157.3$2.50 / $15.002026-03-05
20Claude Haiku 5.5Anthropic
75.1%159.3$0.10 / $0.502026-10-07
21Qwen 3.8 MaxAlibaba
74.7%157.4$2.00 / $6.002026-08-02
22Muse Spark 1.3Meta
74.4%157.2$1.25 / $4.252026-09-02
23GPT-5.2 ProOpenAI
74.0%156.3$21.00 / $1682025-12-11
24Kimi K3Moonshot AI · Open weights
72.2%157.9$3.00 / $15.002026-07-16
25Gemini 3.7 FlashGoogle
71.6%158.0$0.75 / $3.752026-08-13
26Claude Opus 4.7Anthropic
70.2%156.8$5.00 / $25.002026-04-16
27Muse Spark OpenMeta · Open weights
69.8%155.1$1.25 / $4.252026-10-15
28GLM-5.3Z.ai · Open weights
68.8%156.0$1.40 / $4.402026-08-14
29Gemini 3.8 FlashGoogle
68.4%157.4$0.75 / $3.752026-09-02
30GPT-5.2OpenAI
67.4%153.7$1.75 / $14.002025-12-11
31DeepSeek V4.1 FlashDeepSeek · Open weights
67.4%155.9$0.30 / $1.202026-09-09
32Claude Opus 4.6Anthropic
66.0%155.7$5.00 / $25.002026-02-05
32Grok 4.6xAI
66.0%156.6$2.00 / $6.002026-08-12
34Claude Sonnet 5Anthropic
65.6%156.9$2.00 / $10.002026-06-30
34Qwen3.8 Max (0902)Alibaba
65.6%155.1Not measured2026-09-01
36DeepSeek V4 Pro 0813DeepSeek · Open weights
64.6%155.6$1.32 / $3.962026-08-13
36Qwen3.7-MaxAlibaba
64.6%154.0$2.50 / $7.502026-05-19
38Gemini 3.5 FlashGoogle
62.8%154.9$1.50 / $9.002026-05-19
39Gemini 3.1 ProGoogle
59.6%155.3$2.00 / $12.002026-02-19
40GLM-5.2Z.ai · Open weights
59.2%151.9$1.40 / $4.402026-06-16
41Gemini 3.6 FlashGoogle
59.0%154.5$0.75 / $3.752026-07-21
42Mistral Large 4Mistral AI
58.6%150.0$0.68 / $2.092026-10-06
43DeepSeek V4 Flash 0731DeepSeek · Open weights
57.5%154.8$0.19 / $0.382026-07-31
44Grok 4.5xAI
57.2%154.1$2.00 / $6.002026-07-08
44Kimi K2.6Moonshot AI · Open weights
57.2%150.9$0.95 / $4.002026-04-20
46GLM-5.3-FlashZ.ai · Open weights
55.8%153.1$0.15 / $0.502026-08-20
46GPT-5 ProOpenAI
55.8%150.6$15.00 / $1202025-10-07
48GPT-5OpenAI
55.4%150.0$1.25 / $10.002025-08-07
49Kimi K2.7 CodeMoonshot AI · Open weights
54.0%150.1$0.95 / $4.002026-06-12
50Grok 4.7xAI
53.0%154.4$2.00 / $6.002026-09-21
51Gemini 3 FlashGoogle
51.2%151.9$0.50 / $3.002025-12-17
51GPT-5.4 MiniOpenAI
51.2%148.7$0.75 / $4.502026-03-17
53GPT-5 miniOpenAI
46.7%145.3$0.25 / $2.002025-08-07
54Inkling-SmallThinking Machines · Open weights
46.3%150.3$0.47 / $1.202026-07-15
55DeepSeek-V4-ProDeepSeek · Open weights
45.3%149.1$1.32 / $3.962026-04-24
56GPT-5.4 NanoOpenAI
44.9%146.0$0.20 / $1.252026-03-17
56Grok 4.20xAI
44.9%152.1$1.25 / $2.502026-02-17
58Grok 4.3 BetaxAI
42.8%148.8Not measured2026-04-17
59Qwen 3.6 PlusAlibaba
38.3%147.3$0.50 / $3.002026-03-31
60GLM-5.1Z.ai · Open weights
36.8%149.8$1.40 / $4.402026-04-07
61o4-miniOpenAI
36.1%145.5$1.10 / $4.402025-04-16
62Qwen3.6 27BAlibaba · Open weights
35.1%146.2$0.37 / $3.202026-04-22
63Claude Opus 4.5Anthropic
34.4%150.1$5.00 / $25.002025-11-24
63Qwen3.7-PlusAlibaba
34.4%147.0$0.36 / $1.442026-06-02
65InklingThinking Machines · Open weights
33.3%148.4$1.00 / $4.052026-07-15
65o3OpenAI
33.3%146.7$2.00 / $8.002025-04-16
67Qwen3.5 397B-A17BAlibaba · Open weights
31.2%146.8$0.60 / $3.602026-02-13
68Gemini 3.1 Flash-LiteGoogle
27.7%144.3$0.25 / $1.502026-03-03
69GPT-5.5 InstantOpenAI
26.3%142.4Not measured2026-05-05
70Gemini 3.5 Flash-LiteGoogle
26.0%144.9$0.30 / $2.502026-07-21
71Gemini 2.5 Pro (Jun 2025)Google
24.6%145.1$1.25 / $10.002025-06-05
72Claude Sonnet 4.5Anthropic
23.9%146.7$3.00 / $15.002025-09-29
73Qwen 3.6 FlashAlibaba
22.5%142.9Not measured2026-04-27
74Qwen 3.6 35B-A3BAlibaba · Open weights
20.4%143.7$0.20 / $1.102026-04-14
75GPT-5 nanoOpenAI
20.0%139.2$0.05 / $0.402025-08-07
76Qwen3.7 FlashAlibaba
19.3%144.1Not measured2026-07-27
77Qwen3-MaxAlibaba
18.9%142.3$1.66 / $7.222025-09-24
78o3-miniOpenAI
18.6%140.2$1.10 / $4.402025-01-31
79Qwen 3.5 Flash (hosted 35B-A3B)Alibaba
18.3%143.5Not measured2026-02-25
80o1OpenAI
14.7%141.7$15.00 / $60.002024-12-17
81Claude Opus 4.1Anthropic
12.6%144.1$15.00 / $75.002025-08-05
82GPT-4.1 miniOpenAI
6.7%134.9$0.40 / $1.602025-04-14
83GPT-4.1OpenAI
6.0%136.7$2.00 / $8.002025-04-14
84GPT-4 Turbo (Apr 2024)OpenAI
0.7%127.3$10.00 / $30.002024-04-09
84GPT-4o miniOpenAI
0.7%126.5$0.15 / $0.602024-07-18
86GPT-4o (Aug 2024)OpenAI
0.3%128.8$2.50 / $10.002024-08-06
87GPT-3.5 Turbo (Jan 2024)OpenAI
0.0%115.7$0.50 / $1.502024-01-25

How to read this leaderboard

Scores are percentages and higher is better. For each model we show the best result Epoch lists across its effort settings, the same rule for every model. A gap of a point or two is usually within the normal noise of a test, so treat close scores as a tie. Newer models can be missing when nobody has run the test on them yet, so check the release dates. A benchmark measures one kind of skill: use it together with the overall leaderboard rather than on its own.

Questions people ask

What does FrontierMath (Tiers 1-3) measure?

Original, very hard maths problems written by mathematicians. Each has an exact answer and the problems are kept private so models cannot have memorised them. This is the Tiers 1 to 3 set, version 2.

Who ran these tests?

The results come from Epoch AI's open benchmark data. It combines runs by Epoch with results published by other groups, and Epoch's own pages say which is which. We do not run tests ourselves.

Does a high score mean the model is best for me?

Not by itself. FrontierMath (Tiers 1-3) checks one kind of ability. A model can lead here and still be slower, costlier or a poor fit for your work. Compare it on the overall leaderboard and try it on your own tasks.

Why is a model I know missing?

A model appears only when an open source has published results for it. New models can take days or weeks to be measured, and models with no published score are left out rather than guessed.

Where does the data come from?

Scores come from Epoch AI's open benchmark data (licensed CC BY 4.0). We match models across sources, rank them and show them in charts, without changing the scores. Epoch AI does not endorse this site. Prices and context windows come from the open-source LiteLLM price list (MIT licence), which records each model maker's own price page. Prices change often, so confirm on the maker's site before you buy.

How often is this leaderboard updated?

We check our open data sources every day and show the date of the data at the top of the page. A model only moves when its source publishes new results, so a number can stay the same for weeks.

More leaderboards

Different questions need different tests. Pick the one that matches your work.

Where this data comes from

  • Epoch AI: Benchmark data: Epoch AI, 'Capabilities & benchmarking', epoch.ai/benchmarks (CC BY 4.0). We match models across sources, rank them and show them in charts. Scores are not changed. Epoch AI does not endorse this site.
  • LiteLLM model price list: Prices and context windows: LiteLLM model price list (github.com/BerriAI/litellm, MIT licence). We use the model maker's own price entry when there is one. Prices change often, so always confirm on the maker's site before you buy.

Newest data on this page: 26 Oct 2026. Prices are list prices and change often, so confirm on the maker's site before you buy. Sponsors never change a score or an order. How we calculate this · Report a mistake

We use optional cookies for analytics and referral tracking. Learn more about our privacy policy