llmrun
Math

AIME 2024/2025 Leaderboard

AIME (American Invitational Mathematics Examination) is a prestigious high-school competition of hard, integer-answer problems. It's a widely-cited yardstick for multi-step mathematical reasoning, here on the 2024–2025 papers.

Source: epoch34 open models ranked+121 proprietaryData through Jul 2026

Open modelsAll models

Open models ranked on AIME 2024/2025

# shows rank among open models / rank overall (including proprietary).

#ModelScore
1 / 11DeepSeek V4 Pro · 861.6B
96.7%
2 / 12Kimi K2.7 Code · 1058.6B
96.4%
3 / 14Kimi K2.6 · 1058.6B
96.1%
4 / 26GLM 5.1 · 753.9B
92.2%
5 / 28Kimi K2.5 · 1058.6B
92.2%
6 / 34GPT OSS 120B · 120.4B
88.9%
7 / 41Qwen3 235B A22B Thinking 2507 · 235.1B
86.7%
8 / 52GLM 4.7 · 358.3B
83.3%
9 / 53Kimi K2 Thinking · 1058.1B
83.1%
10 / 57GLM 5 · 753.9B
80.0%
11 / 74DeepSeek R1 0528 · 684.5B
66.4%
12 / 86DeepSeek R1 · 684.5B
53.3%
13 / 87DeepSeek R1 Distill Llama 70B · 70.6B
51.4%
14 / 96DeepSeek v3 0324 · 684.5B
37.8%
15 / 103Magistral Small 2506 · 23.6B
30.0%
16 / 108Llama 4 Maverick 17B 128E Instruct · 401.6B
20.6%
17 / 109Gemma 3 27B IT · 27.4B
19.7%
18 / 113DeepSeek v3 · 684.5B
15.8%
19 / 114Phi 4 · 14.7B
13.8%
20 / 116Llama 3.1 405B Instruct · 405.9B
9.7%
21 / 119Qwen2.5 72B Instruct · 72.7B
8.1%
22 / 120Llama 4 Scout 17B 16E Instruct · 108.6B
7.8%
23 / 122Qwen2.5 32B Instruct · 32.8B
7.4%
24 / 133Llama 3.3 70B Instruct · 70.6B
5.1%
25 / 136Llama 3.1 Tulu 3 70B DPO · 70.6B
4.4%
26 / 138Meta Llama 3 70B Instruct · 70.6B
4.3%
27 / 140Llama 3.1 70B Instruct · 70.6B
3.6%
28 / 141Llama 3.2 90B Vision Instruct · 88.6B
2.6%
29 / 144Hermes 2 Theta Llama 3 70B · 70.6B
2.5%
30 / 145Llama 3.1 8B Instruct · 8.0B
2.5%
31 / 149Gemma 2 27B IT · 27.2B
1.4%
32 / 152Meta Llama 3 8B Instruct · 8.0B
0.8%
33 / 153Gemma 2 9B IT · 9.2B
0.6%
34 / 155Llama 2 70B Chat HF · 69.0B
0.0%

Score vs model size

Which models give the most quality for their size — the ones worth running locally.

10B100B1Tmodel size (log scale) →96.7%0.0%Kimi K2.7 Code · 1.1T · 96.4%Kimi K2.6 · 1.1T · 96.1%Kimi K2.5 · 1.1T · 92.2%Qwen3 235B A22B Thinking 2507 · 235B · 86.7%GLM 4.7 · 358B · 83.3%Kimi K2 Thinking · 1.1T · 83.1%GLM 5 · 754B · 80.0%DeepSeek R1 0528 · 685B · 66.4%DeepSeek R1 · 685B · 53.3%DeepSeek v3 0324 · 685B · 37.8%Llama 4 Maverick 17B 128E Instruct · 402B · 20.6%Gemma 3 27B IT · 27B · 19.7%DeepSeek v3 · 685B · 15.8%Llama 3.1 405B Instruct · 406B · 9.7%Qwen2.5 72B Instruct · 73B · 8.1%Llama 4 Scout 17B 16E Instruct · 109B · 7.8%Qwen2.5 32B Instruct · 33B · 7.4%Llama 3.3 70B Instruct · 71B · 5.1%Llama 3.1 Tulu 3 70B DPO · 71B · 4.4%Meta Llama 3 70B Instruct · 71B · 4.3%Llama 3.1 70B Instruct · 71B · 3.6%Llama 3.2 90B Vision Instruct · 89B · 2.6%Hermes 2 Theta Llama 3 70B · 71B · 2.5%Gemma 2 27B IT · 27B · 1.4%Meta Llama 3 8B Instruct · 8B · 0.8%Gemma 2 9B IT · 9B · 0.6%Llama 2 70B Chat HF · 69B · 0.0%Llama 3.1 8B Instruct · 8B · 2.5%Llama 3.1 8B InstructPhi 4 · 15B · 13.8%Phi 4Magistral Small 2506 · 24B · 30.0%Magistral Small 2506DeepSeek R1 Distill Llama 70B · 71B · 51.4%DeepSeek R1 Distill L…GPT OSS 120B · 120B · 88.9%GPT OSS 120BGLM 5.1 · 754B · 92.2%GLM 5.1DeepSeek V4 Pro · 862B · 96.7%DeepSeek V4 Pro
Each dot is a model. Up = higher score, left = smaller (easier to run locally). The dashed line marks the efficiency frontier — the best score you can get at each size or smaller.
  • Llama 3.1 8B Instruct, 8B, score 2.5% — on the efficiency frontier (best score at its size or smaller).
  • Phi 4, 15B, score 13.8% — on the efficiency frontier (best score at its size or smaller).
  • Magistral Small 2506, 24B, score 30.0% — on the efficiency frontier (best score at its size or smaller).
  • DeepSeek R1 Distill Llama 70B, 71B, score 51.4% — on the efficiency frontier (best score at its size or smaller).
  • GPT OSS 120B, 120B, score 88.9% — on the efficiency frontier (best score at its size or smaller).
  • GLM 5.1, 754B, score 92.2% — on the efficiency frontier (best score at its size or smaller).
  • DeepSeek V4 Pro, 862B, score 96.7% — on the efficiency frontier (best score at its size or smaller).

AIME 2024/2025: frequently asked questions

What is the best open LLM on AIME 2024/2025?
DeepSeek V4 Pro is the top open model on AIME 2024/2025, scoring 96.7%. Among all models tested — including proprietary ones — it ranks #11. The top model overall is GPT 5.5 Pro Pre Release (xhigh) (OpenAI) at 100.0%.
What's the best AIME 2024/2025 model you can run on a 24 GB GPU?
Magistral Small 2506 is the highest-scoring open model that fits in 24 GB at 4-bit quantization (about 13 GB), scoring 30.0% on AIME 2024/2025.
What's the best AIME 2024/2025 model you can run on a 12 GB GPU?
Phi 4 is the highest-scoring open model that fits in 12 GB at 4-bit quantization (about 8 GB), scoring 13.8% on AIME 2024/2025.
Can open models match proprietary models on AIME 2024/2025?
Not quite on AIME 2024/2025: the strongest proprietary model (GPT 5.5 Pro Pre Release (xhigh)) scores 100.0%, ahead of the best open model (DeepSeek V4 Pro) at 96.7% — but you can run the open one yourself.

Scores aggregated from epoch. llmrun does not run this benchmark — see the source for methodology, or the about benchmarks for what it measures.