Machine Translation on WMT EN-ZH 24
80COMETkiwiClaude-3.5-Sonnet
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Claude-3.5-Sonnet2025.05 | 80 | — | 86.31 | |
| SSR-X-Zero-7BReasoning/Thinking Mode=true2025.05 | 79.67 | — | 86.75 | |
| Gemini-1.5-Pro2025.05 | 79.54 | — | 87.11 | |
| Qwen3-32BReasoning/Thinking Mode=true2025.05 | 79.51 | — | 86.63 | |
| SSR-Zero-7B2025.05 | 79.34 | — | 85.34 | |
| DeepTrans-7B2025.05 | 78.89 | — | 83.85 | |
| Qwen2.5-32B-Instruct2025.05 | 78.77 | — | 83.48 | |
| Qwen3-8BReasoning/Thinking Mode=true2025.05 | 78.71 | — | 85.31 | |
| Qwen2.5-72B-Instruct2025.05 | 78.18 | — | 85.13 | |
| QwQ-32BReasoning/Thinking Mode=true2025.05 | 78.13 | — | 85.03 | |
| GPT-4o2025.05 | 77.42 | — | 83.95 | |
| Gemma2-27B-it2025.05 | 77.38 | — | 84.17 | |
| Qwen3-32B2025.05 | 76.96 | — | 84.1 | |
| GemmaX2-28-9B-v0.12025.05 | 75.88 | — | 83.33 | |
| Qwen3-8B2025.05 | 75.39 | — | 82.25 | |
| Gemma2-9B-it2025.05 | 74.15 | — | 81.65 | |
| TowerInstruct-13B-v0.12025.05 | 73.81 | — | 80.81 | |
| SSR-Zero-3B2025.05 | 73.57 | — | 79.4 | |
| Qwen2.5-7B-Instruct2025.05 | 72.98 | — | 80.93 | |
| TowerInstruct-7B-v0.22025.05 | 70.87 | — | 77.53 | |
| Qwen2.5-3B-Instruct2025.05 | 67.67 | — | 76.12 | |
| PEGRLBackbone=Qwen3-0.6B, RL training=PEGRL, Rollouts per example=722026.02 | 64.63 | 29.23 | 68.4 | |
| Qwen2.5-7B2025.05 | 64.28 | — | 69.48 | |
| MT-R1-ZeroBackbone=Qwen3-0.6B, RL training=MT-R1-Zero, Rollouts per example=722026.02 | 62.96 | 28.23 | 67.16 | |
| Qwen3-0.6BTraining=Pre-trained only2026.02 | 58.57 | 26.2 | 64.45 | |
| Qwen2.5-3B2025.05 | 23.99 | — | 57.69 |