Evals for Every Language - Language bem: leaderboard
Metric: Average Score (%). Source: huggingface.co. 71 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 66.85 |
| 2 | Claude Opus 4.8 | 60.25 |
| 3 | Gemini 3.1 Flash Lite | 59.46 |
| 4 | Claude Opus 4.7 | 59.05 |
| 5 | GPT-5.4 | 55.63 |
| 6 | Gemini 2.5 Pro | 55.07 |
| 7 | DeepSeek V3.1 | 54.51 |
| 8 | GPT-5.5 | 53.46 |
| 9 | Claude Sonnet 4.6 | 51.32 |
| 10 | Grok 4.20 | 51.31 |
| 11 | Claude Sonnet 4.5 | 50.17 |
| 12 | GPT-5.3 | 47.59 |
| 13 | GPT-5.2 | 47.46 |
| 14 | GPT-5 | 46.96 |
| 15 | GPT-5.1 | 46.39 |
Interactive version: theaggregate.ai/benchmark?slug=evals-for-every-language-language-bem · How It Works · Data refreshed daily, snapshot 2026-09-05.