📐 Math · llm-stats.com
AIME 2025
American Invitational Mathematics Examination problems solved without tool assistance.
9
Modelos testados
Grok 4 Heavy
Modelo principal
100%
Melhor pontuação
xAI
Provedor
| # | Modelo | AIME 2025 |
|---|---|---|
| 1 | Grok 4 Heavy xAI | 100% |
| 2 | Gemini 3 Pro Google | 100% |
| 3 | GPT-5.2 OpenAI | 100% |
| 4 | Claude Opus 4.6 Anthropic | 99.8% |
| 5 | Gemini 3 Flash Google | 99.7% |
| 6 | Seed 2.0 Pro ByteDance | 98.3% |
| 7 | GPT-5 OpenAI | 94.6% |
| 8 | GPT-5.1 OpenAI | 94% |
| 9 | DeepSeek V3.2 DeepSeek | 93.1% |
Dados de benchmark agregados de llm-stats.com. Saiba mais na metodologia.