MIA-Bench — leaderboard
Metric: Overall Score. Source: huggingface.co. 13 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 (2025-04-14) | 94.55 |
| 2 | Grok 2 (1212) | 93.61 |
| 3 | Gemini 2.0 Flash | 92.93 |
| 4 | InternVL3-78B | 92.65 |
| 5 | Gemini 2.5 Flash | 92.17 |
| 6 | InternVL3-38B | 92.05 |
| 7 | InternVL2.5-78B | 91.97 |
| 8 | Llama 4 Scout Instruct | 91.57 |
| 9 | Claude 3.5 Sonnet (20241022) | 91.46 |
| 10 | Pixtral-12B | 90.28 |
| 11 | Claude 3.7 Sonnet | 89.9 |
| 12 | InternVL3-8B | 89.7 |
| 13 | Llama 3.2 11B Instruct | 76.57 |
Interactive version: aibenchmarks.dev/benchmark?slug=mia-bench · How the rankings work · Data refreshed daily, snapshot 2026-07-13.