Translation instruction-following on IFMTBench
91.95Simple ScoreGemini 3.1 Pro
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Gemini 3.1 ProThinking mode=true2026.05 | 91.95 | 84.53 | 89.08 | |
| Hy-MT2-30B-A3BThinking mode=false2026.05 | 90.2 | 75.94 | 84.69 | |
| Hy-MT2-7BThinking mode=false2026.05 | 89.73 | 72.67 | 83.14 | |
| GPT-5.5Thinking mode=true2026.05 | 89.25 | 84.44 | 87.39 | |
| GLM5.1Thinking mode=true2026.05 | 88.81 | 82.14 | 86.23 | |
| Kimi K2.6Thinking mode=false2026.05 | 88.07 | 79.92 | 84.92 | |
| GPT-5.5Thinking mode=false2026.05 | 86.97 | 83.74 | 85.72 | |
| DeepSeek-V4-ProThinking mode=true2026.05 | 86.97 | 83.74 | 85.72 | |
| GLM5.1Thinking mode=false2026.05 | 85.53 | 78.83 | 82.94 | |
| Qwen3.5-397B-A17BThinking mode=true2026.05 | 85.05 | 72.08 | 80.04 | |
| Kimi K2.6Thinking mode=true2026.05 | 83.17 | 80.81 | 82.26 | |
| Gemma4-26B-A4BThinking mode=false2026.05 | 83.02 | 73.18 | 79.22 | |
| Qwen3.5-397B-A17BThinking mode=false2026.05 | 81.71 | 77.61 | 80.13 | |
| DeepSeek-V4-ProThinking mode=false2026.05 | 81.38 | 78.23 | 80.16 | |
| Qwen3.6-35B-A3BThinking mode=false2026.05 | 77.79 | 69.7 | 74.66 | |
| Hy-MT2-1.8BThinking mode=false2026.05 | 76.76 | 57.61 | 69.36 | |
| Gemma4-E4BThinking mode=false2026.05 | 74.67 | 66.98 | 71.7 | |
| Gemma4-E2BThinking mode=false2026.05 | 62.93 | 49.1 | 57.59 |