Evals for Every Language - Language as: leaderboard
Metric: Average Score (%). Source: huggingface.co. 71 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 69.38 |
| 2 | Claude Opus 4.7 | 68.11 |
| 3 | GPT-5.3 | 67.23 |
| 4 | Claude Opus 4.8 | 66.93 |
| 5 | Gemini 2.5 Pro | 66.33 |
| 6 | Qwen 3.6 Plus | 66.13 |
| 7 | GPT-5 | 66.04 |
| 8 | GPT-5.5 | 66.04 |
| 9 | Claude Sonnet 4.6 | 65.28 |
| 10 | GPT-5.2 | 64.84 |
| 11 | DeepSeek V3.1 | 64.63 |
| 12 | GPT-5.1 | 64.43 |
| 13 | GPT-5.4 | 64.14 |
| 14 | Grok 4.20 | 64.04 |
| 15 | Claude Sonnet 4 | 62.62 |
Interactive version: theaggregate.ai/benchmark?slug=evals-for-every-language-language-as · How It Works · Data refreshed daily, snapshot 2026-09-05.