Evals for Every Language - Language bug: leaderboard
Metric: Average Score (%). Source: huggingface.co. 70 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 44.81 |
| 2 | Claude Opus 4.8 | 42.09 |
| 3 | Claude Opus 4.7 | 41.91 |
| 4 | Gemini 3.1 Flash Lite | 40.71 |
| 5 | Gemini 2.5 Pro | 39.16 |
| 6 | GPT-5.5 | 38.64 |
| 7 | GPT-5.1 | 38.55 |
| 8 | Claude Sonnet 4 | 38.12 |
| 9 | Claude Sonnet 4.6 | 37.26 |
| 10 | Claude Sonnet 4.5 | 36.84 |
| 11 | GPT-5.3 | 36.52 |
| 12 | DeepSeek V3.1 | 35.55 |
| 13 | DeepSeek V3.2 Exp | 35.31 |
| 14 | GPT-5.4 | 35.25 |
| 15 | GPT-5 | 35.15 |
Interactive version: theaggregate.ai/benchmark?slug=evals-for-every-language-language-bug · How It Works · Data refreshed daily, snapshot 2026-09-05.