Evals for Every Language - Translation From: leaderboard
LanguageBench Translation From spBLEU averaged across available languages and scaled to 0-100. Translation uses spBLEU only, without mixing chrF; higher is better.
Metric: Average Score (%). Source: huggingface.co. Saturation forecast: Around 2029. 71 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 32.83 | #58 |
| 2 | Gemini 2.5 Pro | 31.79 | #152 |
| 3 | Gemini 2.5 Flash | 30.36 | #245 |
| 4 | Claude Sonnet 4.5 | 29.35 | #142 |
| 5 | Claude Opus 4.7 | 28.26 | #49 |
| 6 | GPT-5.5 | 28.01 | #31 |
| 7 | Claude Opus 4.8 | 27.42 | #37 |
| 8 | Gemini 3.1 Flash Lite | 27.42 | #191 |
| 9 | Gemini 2.5 Flash Lite | 26.74 | #422 |
| 10 | Qwen 3.6 Plus | 26.22 | #120 |
| 11 | Claude Sonnet 4.6 | 26.19 | #90 |
| 12 | GPT-5.4 | 25.51 | #79 |
| 13 | Claude Sonnet 4 | 25.09 | #199 |
| 14 | GPT-5 | 24.87 | #98 |
| 15 | GPT-5.1 | 24.62 | #135 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=evals-for-every-language-translation-from · How It Works · Data refreshed daily, snapshot 2026-10-20.