벤치마크 · 수학
FrontierMath 4단계(2025판) 순위
FrontierMath 이전 판의 4단계 기록. 2026년 판과 분리해 표시한다.
상위 모델
- OpenAI
- Anthropic
Google- xAI
공개일에 따른 점수
선은 개발사별로 그때까지의 최고 점수를 잇습니다. 오른쪽 위로 갈수록 최근에 나온 높은 점수입니다.
- GPT-5.5 Pro 40%
- Claude Opus 4.8 31%
- Gemini 3 Pro 19%
- Grok 4 2%
전체 순위
| 순위 | 모델 | 점수 | 생각 수준 | 공개일 |
|---|---|---|---|---|
| 1 | GPT-5.5 ProOpenAI | 39.6%±7.1% | 매우 높음 | 2026-04-23 |
| 2 | GPT-5.4 ProOpenAI | 37.5%±7% | — | 2026-03-05 |
| 3 | GPT-5.5OpenAI | 35.4%±6.9% | 매우 높음 | 2026-04-23 |
| 4 | GPT-5.2 ProOpenAI | 31.3%±6.7% | — | 2025-12-11 |
| 5 | Claude Opus 4.8Anthropic | 31.3%±6.8% | — | 2026-05-28 |
| 6 | GPT-5.4OpenAI | 27.1%±6.4% | 매우 높음 | 2026-03-05 |
| 7 | Claude Opus 4.7Anthropic | 22.9%±6.1% | 매우 높음 | 2026-04-16 |
| 8 | Claude Opus 4.6Anthropic | 22.9%±6.1% | 최대 | 2026-02-05 |
| 9 | GPT-5.2OpenAI | 18.8%±5.6% | 매우 높음 | 2025-12-11 |
| 10 | 18.8%±5.7% | — | 2025-11-18 | |
| 11 | 16.7%±5.4% | — | 2026-02-19 | |
| 12 | GPT-5 ProOpenAI | 14.6%±5.1% | — | 2025-10-07 |
| 13 | 14.6%±5.1% | — | 2026-05-19 | |
| 14 | GPT-5.1OpenAI | 12.5%±4.8% | 높음 | 2025-11-13 |
| 14 | GPT-5OpenAI | 12.5%±4.8% | 높음 | 2025-08-07 |
| 16 | Claude Sonnet 4.6Anthropic | 8.3%±4% | — | 2026-02-17 |
| 17 | GPT-5.4 NanoOpenAI | 6.3%±3.5% | 높음 | 2026-03-17 |
| 17 | GPT-5 miniOpenAI | 6.3%±3.5% | 높음 | 2025-08-07 |
| 17 | o4-miniOpenAI | 6.3%±3.5% | 높음 | 2025-04-16 |
| 20 | 4.2%±2.9% | — | 2025-12-17 | |
| 20 | Claude Opus 4.5Anthropic | 4.2%±2.9% | — | 2025-11-24 |
| 20 | Claude Sonnet 4.5Anthropic | 4.2%±2.9% | — | 2025-09-29 |
| 20 | Claude Opus 4.1Anthropic | 4.2%±2.9% | — | 2025-08-05 |
| 20 | 4.2%±2.9% | — | 2025-06-17 | |
| 20 | 4.2%±2.9% | — | 2025-06-05 | |
| 20 | Claude Opus 4Anthropic | 4.2%±2.9% | — | 2025-05-22 |
| 20 | o3-miniOpenAI | 4.2%±2.9% | 높음 | 2025-01-31 |
| 28 | GPT-5.4 MiniOpenAI | 2.1%±2.1% | 높음 | 2026-03-17 |
| 28 | Claude Haiku 4.5Anthropic | 2.1%±2.1% | — | 2025-10-15 |
| 28 | GPT-5 nanoOpenAI | 2.1%±2.1% | 중간 | 2025-08-07 |
| 28 | Grok 4 HeavyxAI | 2.1%±2.1% | — | 2025-07-10 |
| 28 | Grok 4xAI | 2.1%±2.1% | — | 2025-07-09 |
| 28 | o3OpenAI | 2.1%±2.1% | 높음 | 2025-04-16 |
| 34 | Claude Sonnet 4Anthropic | 0% | — | 2025-05-22 |
| 34 | GPT-4.1OpenAI | 0% | — | 2025-04-14 |
| 34 | Grok 3xAI | 0% | — | 2025-04-09 |
| 34 | Claude 3.5 Sonnet (October 2024)Anthropic | 0% | — | 2024-10-22 |
| 34 | Claude 3.5 SonnetAnthropic | 0% | — | 2024-06-20 |