OlympicArena — leaderboard
Metric: Overall Accuracy (%). Source: gair-nlp.github.io. 28 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4o | 34.01 |
| 2 | Claude 3.5 Sonnet | 33.54 |
| 3 | DeepSeek Coder V2 | 29.31 |
| 4 | Qwen 2 72B Instruct | 28.84 |
| 5 | Gemini 1.5 Pro | 28.68 |
| 6 | GPT-4o Mini | 28.37 |
| 7 | Llama 3 70B Instruct | 21.63 |
| 8 | Qwen 1.5 32B Chat | 21.16 |
| 9 | Gemma 2 27B (IT) | 19.91 |
| 10 | Claude 3 Sonnet | 18.97 |
| 11 | internlm2.5-7B-chat | 18.65 |
| 12 | Llama 3.1 8B Instruct | 14.73 |
| 13 | InternVL2-8B | 14.73 |
| 14 | internlm2-chat-20B | 13.95 |
| 15 | Yi 34B (Chat) | 12.54 |
Interactive version: theaggregate.ai/benchmark?slug=olympicarena · How the rankings work · Data refreshed daily, snapshot 2026-07-22.