SeaEval - FLORES Translation - Chinese-to-English (Zero-Shot): leaderboard
SeaEval zero-shot Chinese-to-English translation evaluation measuring BLEU quality for translation into English.
Metric: BLEU (0-100). Source: huggingface.co. Saturation forecast: Estimated already saturated. 24 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 72B Instruct | 28.43 | #454 |
| 2 | Llama 3.1 70B Instruct | 28.33 | #561 |
| 3 | Llama 3.1 70B Cpt Sea Lionv3 Instruct | 27.86 | #548 |
| 4 | Qwen 2.5 32B Instruct | 26.92 | #498 |
| 5 | Gemma 2 9B (IT) | 26.75 | #788 |
| 6 | Qwen 2.5 14B Instruct | 26.28 | #647 |
| 7 | SeaLLMs-v3-7B-Chat | 25.17 | #915 |
| 8 | gemma2-9B-cpt-sea-lionv3-instruct | 25.11 | #705 |
| 9 | Llama 3 70B Instruct | 24.4 | #634 |
| 10 | Qwen 2.5 7B Instruct | 24.37 | #862 |
| 11 | Llama 3.1 8B Instruct | 24.29 | #1035 |
| 12 | Qwen 2 72B Instruct | 23.89 | #576 |
| 13 | Llama 3 8B Cpt Sea Lionv2 Instruct | 23.82 | #1188 |
| 14 | Llama 3 8B Cpt Sea Lionv2.1 Instruct | 23.78 | #1024 |
| 15 | Qwen 2.5 3B Instruct | 22.45 | #1162 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=seaeval-flores-translation-chinese-to-english-zero-shot · How It Works · Data refreshed daily, snapshot 2026-10-20.