Evals for Every Language - Language mag: leaderboard
Metric: Average Score (%). Source: huggingface.co. 70 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 59.83 |
| 2 | Gemini 2.5 Flash | 59.48 |
| 3 | Gemini 2.5 Pro | 59.42 |
| 4 | Claude Sonnet 4.5 | 59.17 |
| 5 | Claude Opus 4.7 | 58.53 |
| 6 | Gemini 3.1 Flash Lite | 58.12 |
| 7 | Claude Opus 4.8 | 57.79 |
| 8 | GPT-5.4 | 56.59 |
| 9 | Claude Sonnet 4 | 56.24 |
| 10 | GPT-5.3 | 56.2 |
| 11 | GPT-5.2 | 56.12 |
| 12 | Nova Pro (v1) | 55.81 |
| 13 | GPT-5.1 | 55.57 |
| 14 | GPT-5.5 | 53.6 |
| 15 | Claude Sonnet 4.6 | 52.97 |
Interactive version: theaggregate.ai/benchmark?slug=evals-for-every-language-language-mag · How It Works · Data refreshed daily, snapshot 2026-09-05.