AIME (Vals)
Recent AIME problems, run by Vals AI at stated reasoning effort.
- Published by
- Vals AI
- Category
- Maths
- Index weight
- 0.5
- Models
- 91
- Data as of
- 9 Sept 2026
Vals AI (vals.ai).
- 1Gemini 3.1 Pro98.1%
- 2Muse Spark96.9%
- 3GPT-5.296.9%
- 4Gemini 3 Pro96.7%
- 5GPT-5.496.7%
- 6Grok 4.2096.5%
- 7Claude Opus 4.796.3%
- 8Claude Opus 4.695.6%
- 9Gemini 3 Flash95.6%
- 10Kimi K2.595.6%
- 11GPT-5.4 mini95.6%
- 12Claude Opus 4.595.4%
- 13Qwen3.6 Plus94.6%
- 14GPT-593.4%
- 15GLM-4.793.3%
91 of 91
| # | |||||
|---|---|---|---|---|---|
| 1 |  Gemini 3.1 ProGoogle | 98.1% | – | 60.7 | – |
| 2 |  Muse SparkMeta | 96.9% | – | 65.4 | – |
| 3 |  GPT-5.2OpenAI | 96.9% | – | 57.3 | – |
| 4 | Gemini 3 ProGoogle | 96.7% | – | 57.0 | – |
| 5 | GPT-5.4OpenAI | 96.7% | – | 61.3 | – |
| 6 |  Grok 4.20xAI | 96.5% | – | 54.3 | – |
| 7 |  Claude Opus 4.7Anthropic | 96.3% | – | 65.5 | – |
| 8 | Claude Opus 4.6Anthropic | 95.6% | – | 62.3 | – |
| 9 | Gemini 3 FlashGoogle | 95.6% | – | 58.5 | – |
| 10 |  Kimi K2.5Moonshot AIopen | 95.6% | – | 55.3 | – |
| 11 | GPT-5.4 miniOpenAI | 95.6% | – | 46.0 | – |
| 12 | Claude Opus 4.5Anthropic | 95.4% | – | 60.6 | – |
| 13 |  Qwen3.6 PlusAlibabaopen | 94.6% | – | 59.0 | – |
| 14 | GPT-5OpenAI | 93.4% | – | 54.5 | – |
| 15 |  GLM-4.7Zhipu AIopen | 93.3% | zai/glm-4.7 | 54.9 | – |
| 16 | GPT-5.1OpenAI | 93.3% | – | 53.8 | – |
| 17 | GLM-4.6Zhipu AIopen | 92.7% | zai/glm-4.6 | 47.6 | – |
| 18 | gpt-oss-120bOpenAIopen | 92.6% | – | 47.6 | – |
| 19 | Qwen3.5 FlashAlibaba | 92.5% | – | 51.8 | – |
| 20 | Claude Sonnet 4.6Anthropic | 92.3% | – | 60.3 | – |
| 21 | GLM-5.1Zhipu AIopen | 91.9% | zai/glm-5.1 | 60.1 | – |
| 22 | Grok 4.1xAI | 91.9% | – | 55.9 | – |
| 23 | GLM-5Zhipu AIopen | 91.7% | – | 57.0 | – |
| 24 | GPT-5 miniOpenAI | 91.5% | – | 52.3 | – |
| 25 | Grok 4 FastxAI | 91.3% | – | 53.0 | – |
| 26 |  MiniMax-M2.7MiniMaxopen | 91.0% | minimax/MiniMax-M2.7 | 56.5 | – |
| 27 | Grok 4xAI | 90.6% | 0709 | 57.2 | – |
| 28 | MiniMax-M2.5MiniMaxopen | 88.8% | minimax/MiniMax-M2.5 | 53.5 | – |
| 29 | GPT-5.4 nanoOpenAI | 88.8% | – | 48.3 | – |
| 30 | Claude Sonnet 4.5Anthropic | 88.2% | – | 52.8 | – |
| 31 | GLM-4.5Zhipu AIopen | 86.7% | zai/glm-4.5 | 50.2 | – |
| 32 | o3-miniOpenAI | 86.5% | – | 47.8 | – |
| 33 | Qwen3.5 PlusAlibaba | 86.0% | – | 57.2 | – |
| 34 | gpt-oss-20bOpenAIopen | 86.0% | – | 47.1 | – |
| 35 | Gemini 2.5 ProGoogle | 85.8% | – | 54.1 | – |
| 36 | Kimi K2 ThinkingMoonshot AIopen | 85.4% | – | 52.8 | – |
| 37 | o3OpenAI | 85.3% | – | 52.9 | – |
| 38 | Grok 3 Mini FastxAI | 85.0% | – | 53.5 | – |
| 39 |  DeepSeek V3.2DeepSeekopen | 84.6% | – | 50.1 | – |
| 40 | Qwen3 235B A22BAlibabaopen | 84.0% | – | 48.7 | – |
| 41 | o4-miniOpenAI | 83.7% | – | 48.0 | – |
| 42 |  Magistral MediumMistral AIopen | 83.5% | medium, 2509 | 44.1 | – |
| 43 | Gemini 3.1 Flash LiteGoogle | 83.3% | – | 49.1 | – |
| 44 | Claude Haiku 4.5Anthropic | 82.7% | – | 48.0 | – |
| 45 | GPT-5 nanoOpenAI | 81.2% | – | 45.9 | – |
| 46 | Qwen3 MaxAlibaba | 81.0% | – | 52.7 | – |
| 47 | Magistral SmallMistral AIopen | 80.7% | 2509 | 40.5 | – |
| 48 | Claude Opus 4.1Anthropic | 78.2% | – | 58.1 | – |
| 49 | MiniMax-M2MiniMaxopen | 77.9% | minimax/MiniMax-M2.1 | 55.1 | – |
| 50 | Claude Opus 4.5Anthropic | 76.9% | – | 58.0 | – |
| 51 | Claude Sonnet 4Anthropic | 76.3% | – | 54.0 | – |
| 52 | DeepSeek R1DeepSeekopen | 74.0% | fireworks/deepseek-r1 | 48.9 | – |
| 53 | o1OpenAI | 71.5% | – | 44.3 | – |
| 54 | Grok 3 Mini FastxAI | 70.6% | – | 50.3 | – |
| 55 | Kimi K2Moonshot AIopen | 62.7% | – | 48.8 | – |
| 56 | Qwen3 MaxAlibaba | 60.7% | – | 51.8 | – |
| 57 | Grok 3xAI | 58.8% | grok/grok-3 | 50.0 | – |
| 58 | Langston Nim Nvidia Llama 3.3 Nemotron Super 49B v1 42e84561Unknown | 53.5% | – | – | – |
| 59 | Gemini 2.5 Flash 09Google | 51.5% | thinking, preview, 2025 | 52.5 | – |
| 60 | Gemini 2.5 Flash 09Google | 49.8% | preview, 2025 | 50.7 | – |