BENCHMARK · 수학
OTIS 모의 AIME 순위
OTIS의 경시대회형 수학 문제 45개. MATH Level 5보다 어렵고 FrontierMath보다 쉽다.
상위 모델 여러 개가 95% 이상을 받아 이 시험은 최신 모델끼리의 차이를 가리기 어려워졌습니다(포화). 최신 모델 비교에는 다른 벤치마크를 함께 보세요.
상위 모델
- OpenAI
- Anthropic
Google- xAI
세로 점선은 무작위로 찍었을 때 기대되는 점수입니다.
공개일에 따른 점수
선은 개발사별로 그때까지의 최고 점수를 잇습니다. 오른쪽 위로 갈수록 최근에 나온 높은 점수입니다.
- GPT-5.5 Pro 100%
- Claude Fable 5 100%
- Grok 4.6 99%
- Gemini 3.8 Flash 99%
전체 순위
| 순위 | 모델 | 점수 | 생각 수준 | 공개일 |
|---|---|---|---|---|
| 1 | GPT-6.1 SolOpenAI | 100% | 최대 | 2026-09-29 |
| 1 | Claude Sonnet 5.5Anthropic | 100% | 최대 | 2026-09-28 |
| 1 | GPT-6 SolOpenAI | 100% | 최대 | 2026-09-22 |
| 1 | Claude Opus 5.5Anthropic | 100% | 최대 | 2026-09-22 |
| 1 | GPT-6 AstraOpenAI | 100% | 최대 | 2026-09-03 |
| 1 | Claude Fable 5.1Anthropic | 100% | 최대 | 2026-09-01 |
| 1 | GPT-5.6 SolOpenAI | 100% | 최대 | 2026-07-09 |
| 1 | Claude Fable 5Anthropic | 100% | 높음 | 2026-06-09 |
| 1 | GPT-5.5 ProOpenAI | 100% | 매우 높음 | 2026-04-23 |
| 1 | GPT-5.5OpenAI | 100% | 매우 높음 | 2026-04-23 |
| 11 | GPT-5.6 TerraOpenAI | 99.7%±0.3% | 최대 | 2026-07-09 |
| 12 | Grok 4.6xAI | 99.2%±0.5% | 매우 높음 | 2026-08-12 |
| 13 | Claude Haiku 5.5Anthropic | 98.9%±0.5% | 매우 높음 | 2026-10-07 |
| 13 | GPT-6 LunaOpenAI | 98.9%±0.7% | 최대 | 2026-09-22 |
| 13 | Gemini 3.8 FlashGoogle | 98.9%±0.9% | 높음 | 2026-09-02 |
| 13 | Claude Opus 5Anthropic | 98.9%±1.1% | 최대 | 2026-07-24 |
| 17 | GPT-5.6 LunaOpenAI | 98.3%±1.2% | 최대 | 2026-07-09 |
| 17 | Claude Opus 4.8Anthropic | 98.3%±1.4% | — | 2026-05-28 |
| 19 | Grok 4.7xAI | 98.1%±1.4% | 매우 높음 | 2026-09-21 |
| 20 | Claude Opus 4.7Anthropic | 97.8%±2.2% | 매우 높음 | 2026-04-16 |
| 21 | Grok 4.5xAI | 97.8%±1.3% | 높음 | 2026-07-08 |
| 21 | GPT-5.4OpenAI | 97.8%±2.2% | 높음 | 2026-03-05 |
| 23 | Gemini 3.7 FlashGoogle | 97.2%±1.9% | 높음 | 2026-08-13 |
| 24 | GPT-5.2OpenAI | 96.1%±2.6% | 높음 | 2025-12-11 |
| 25 | Gemini 3.1 ProGoogle | 95.6%±3.1% | — | 2026-02-19 |
| 26 | Gemini 3.5 FlashGoogle | 95.6%±2.7% | — | 2026-05-19 |
| 26 | Gemini 3 FlashGoogle | 95.6%±3.1% | — | 2025-12-17 |
| 28 | Claude Sonnet 5Anthropic | 94.7%±2.3% | 매우 높음 | 2026-06-30 |
| 29 | Claude Opus 4.6Anthropic | 94.4%±2.8% | — | 2026-02-05 |
| 30 | Gemini 3.6 FlashGoogle | 94.2%±3.1% | — | 2026-07-21 |
| 31 | Grok 4.3 BetaxAI | 93.3%±3% | — | 2026-04-17 |
| 32 | Grok 4.20xAI | 92.2%±3.3% | — | 2026-02-17 |
| 33 | Gemini 3 ProGoogle | 91.4%±3.7% | — | 2025-11-18 |
| 33 | GPT-5OpenAI | 91.4%±3.7% | 높음 | 2025-08-07 |
| 35 | GPT-5.4 MiniOpenAI | 88.9%±4.7% | 매우 높음 | 2026-03-17 |
| 35 | gpt-oss-120bOpenAI | 88.9%±4.5% | — | 2025-08-05 |
| 37 | GPT-5.1OpenAI | 88.6%±4% | 높음 | 2025-11-13 |
| 38 | GPT-5.4 NanoOpenAI | 87.8%±3.9% | 높음 | 2026-03-17 |
| 39 | GPT-5 miniOpenAI | 86.7%±4.3% | 높음 | 2025-08-07 |
| 40 | Claude Opus 4.5Anthropic | 86.1%±4.3% | — | 2025-11-24 |
| 41 | Claude Sonnet 4.6Anthropic | 85.8%±4.5% | — | 2026-02-17 |
| 42 | Gemini 2.5 Pro (Jun 2025)Google | 84.7%±4.8% | — | 2025-06-05 |
| 43 | o3OpenAI | 84.4%±5.5% | 중간 | 2025-04-16 |
| 44 | Grok 4xAI | 84%±5% | — | 2025-07-09 |
| 45 | o4-miniOpenAI | 81.7%±4.7% | 높음 | 2025-04-16 |
| 46 | GPT-5 nanoOpenAI | 81.1%±4.6% | 높음 | 2025-08-07 |
| 47 | Gemini 3.1 Flash-LiteGoogle | 80%±6% | — | 2026-03-03 |
| 48 | Claude Sonnet 4.5Anthropic | 77.8%±6.3% | — | 2025-09-29 |
| 48 | Grok-3 minixAI | 77.8%±6.3% | — | 2025-06-24 |
| 50 | o3-miniOpenAI | 76.9%±5.2% | 높음 | 2025-01-31 |
| 51 | o1OpenAI | 73.3%±6.7% | 높음 | 2024-12-17 |
| 52 | Gemini 2.5 Flash (Apr 2025)Google | 73.1%±5.5% | — | 2025-04-17 |
| 53 | Gemini 3.5 Flash-LiteGoogle | 71.1%±6.8% | — | 2026-07-21 |
| 53 | Claude Sonnet 4Anthropic | 71.1%±6.8% | — | 2025-05-22 |
| 55 | Gemini 2.5 Flash (May 2025)Google | 70.8%±5.7% | — | 2025-05-20 |
| 56 | Claude Opus 4.1Anthropic | 68.9%±7% | — | 2025-08-05 |
| 57 | GPT-5.5 InstantOpenAI | 68.1%±6.2% | — | 2026-05-05 |
| 58 | Claude Haiku 4.5Anthropic | 66.7%±7.1% | — | 2025-10-15 |
| 59 | gpt-oss-20bOpenAI | 65.3%±5.9% | — | 2025-08-05 |
| 60 | Claude Opus 4Anthropic | 64.4%±7.2% | — | 2025-05-22 |
| 61 | Claude 3.7 SonnetAnthropic | 57.8%±7.4% | — | 2025-02-24 |
| 62 | Grok 3xAI | 55.6%±7.5% | — | 2025-04-09 |
| 63 | o1-miniOpenAI | 46.9%±6.5% | 높음 | 2024-09-12 |
| 64 | GPT-4.1 miniOpenAI | 44.7%±6.2% | — | 2025-04-14 |
| 65 | GPT-4.1OpenAI | 38.3%±6.3% | — | 2025-04-14 |
| 66 | GPT-4.5OpenAI | 37.8%±7.3% | — | 2025-02-27 |
| 67 | o1-previewOpenAI | 31.1%±7% | — | 2024-09-12 |
| 68 | GPT-4.1 nanoOpenAI | 28.9%±6% | — | 2025-04-14 |
| 69 | Gemini 1.5 Pro (Sept 2024)Google | 23.1%±5.2% | — | 2024-09-24 |
| 70 | Gemini 1.5 Flash (Sep 2024)Google | 16.3%±4.1% | — | 2024-09-24 |
| 71 | Grok-2 (Dec 2024)xAI | 11.5%±3.4% | — | 2024-12-12 |
| 72 | Claude 3.5 Sonnet (October 2024)Anthropic | 8.5%±2.6% | — | 2024-10-22 |
| 73 | GPT-4o miniOpenAI | 6.9%±2.7% | — | 2024-07-18 |
| 74 | Gemini 1.5 Pro (May 2024)Google | 6.8%±1.9% | — | 2024-05-14 |
| 75 | GPT-4 Turbo (Apr 2024)OpenAI | 6.7%±2.4% | — | 2024-04-09 |
| 76 | Claude 3.5 SonnetAnthropic | 6.5%±2.2% | — | 2024-06-20 |
| 77 | GPT-4o (Aug 2024)OpenAI | 6.4%±2.6% | — | 2024-08-06 |
| 78 | GPT-4o (Nov 2024)OpenAI | 6.3%±2.4% | — | 2024-11-20 |
| 78 | GPT-4o (May 2024)OpenAI | 6.3%±2.3% | — | 2024-05-13 |
| 80 | Claude 3 OpusAnthropic | 4.7%±2% | — | 2024-02-29 |
| 81 | Gemini 1.5 Flash 8BGoogle | 4.6%±2% | — | 2024-10-03 |
| 82 | Claude 3.5 HaikuAnthropic | 4.3%±1.6% | — | 2024-10-22 |
| 83 | Gemini 1.5 Flash (May 2024)Google | 3.9%±1.4% | — | 2024-05-23 |
| 84 | Claude 3 SonnetAnthropic | 2.5%±1.9% | — | 2024-02-29 |
| 84 | Claude 2Anthropic | 2.5%±1.5% | — | 2023-07-11 |
| 86 | GPT-3.5 Turbo (Jan 2024)OpenAI | 2.2%±2.2% | — | 2024-01-25 |
| 87 | Claude 2.1Anthropic | 1.9%±1.1% | — | 2023-11-21 |
| 88 | Claude 3 HaikuAnthropic | 1.8%±0.8% | — | 2024-03-07 |
| 89 | Gemini 1.0 ProGoogle | 1.1%±0.6% | — | 2023-12-13 |
| 89 | GPT-4 (Jun 2023)OpenAI | 1.1%±0.5% | — | 2023-06-13 |
| 91 | GPT-4 (Mar 2023)OpenAI | 0.6%±0.4% | — | 2023-03-14 |
다른 벤치마크
- GPQA Diamond 과학
- MATH Level 5 수학
- FrontierMath(2025판) 수학
- FrontierMath 1~3단계 수학
- FrontierMath 4단계 수학
- SWE-bench Verified 코딩
- MirrorCode 코딩
- SimpleQA Verified 지식
- 체스 퍼즐 추론
- 미공개 게임 퍼즐 추론
- 가구 조립 검사 공간·학습
- EBR-bench 공간·학습
- Aider Polyglot 코딩