Evals for Every Language - Language ay: leaderboard
Metric: Average Score (%). Source: huggingface.co. 71 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 62.7 |
| 2 | Claude Opus 4.7 | 59.38 |
| 3 | GPT-5.5 | 59.02 |
| 4 | DeepSeek V3.1 | 58.91 |
| 5 | Claude Opus 4.8 | 58.4 |
| 6 | Claude Sonnet 4.6 | 57.7 |
| 7 | Gemini 3.1 Flash Lite | 57.64 |
| 8 | GPT-5.4 | 57.15 |
| 9 | GPT-5.2 | 57.05 |
| 10 | GPT-5.3 | 54.63 |
| 11 | Grok 4.20 | 54.34 |
| 12 | GPT-5 | 54.01 |
| 13 | Qwen 3.6 Plus | 53.13 |
| 14 | Gemini 2.5 Pro | 52.35 |
| 15 | Claude Sonnet 4.5 | 51.74 |
Interactive version: theaggregate.ai/benchmark?slug=evals-for-every-language-language-ay · How It Works · Data refreshed daily, snapshot 2026-09-05.