MixEval: leaderboard
Derives queries from real-world user interactions and matches them with similar existing benchmark queries. Achieves 0.96 correlation with Chatbot Arena while being much cheaper to run.
Metric: Score. Source: mixeval.github.io. Saturation forecast: Estimated already saturated. 52 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | O1 Preview | 72 | #193 |
| 2 | Llama 3.1 405B Instruct | 66.2 | #457 |
| 3 | GPT-4o (2024-05-13) | 64.7 | #325 |
| 4 | Claude 3 Opus | 63.5 | #571 |
| 5 | GPT-4 Turbo | 62.6 | #404 |
| 6 | Mistral Large 2 (Jul) | 57.4 | #484 |
| 7 | Llama 3 70B Instruct | 55.9 | #634 |
| 8 | Claude 3 Sonnet | 54 | #775 |
| 9 | DeepSeek V2 | 51.7 | #549 |
| 10 | GPT-4o Mini | 51.6 | #599 |
| 11 | Command-R+ | 51.4 | #891 |
| 12 | Yi 1.5 34B Chat | 51.2 | #800 |
| 13 | Mistral Large | 50.3 | #710 |
| 14 | Qwen 1.5 72B Chat | 48.3 | #776 |
| 15 | Mistral Medium | 47.8 | #792 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=mixeval · How It Works · Data refreshed daily, snapshot 2026-10-20.