Multi-turn instruction-following on Multi-IF
95.02Turn 1 ScoreGemini 3.1 Pro
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Gemini 3.1 ProThinking mode=true2026.05 | 95.02 | 89.57 | 84.89 | |
| GPT-5.5Thinking mode=true2026.05 | 94.9 | 89.33 | 84.46 | |
| GLM5.1Thinking mode=true2026.05 | 93.94 | 86.47 | 82.77 | |
| Kimi K2.6Thinking mode=true2026.05 | 93.87 | 87.15 | 82.29 | |
| DeepSeek-V4-ProThinking mode=true2026.05 | 93.62 | 87.02 | 81.69 | |
| GPT-5.5Thinking mode=false2026.05 | 93.25 | 87.21 | 82.02 | |
| GLM5.1Thinking mode=false2026.05 | 92.24 | 85.25 | 79.67 | |
| Qwen3.5-397B-A17BThinking mode=true2026.05 | 90.81 | 84.17 | 79.56 | |
| Kimi K2.6Thinking mode=false2026.05 | 90.6 | 84.93 | 77.48 | |
| Qwen3.5-397B-A17BThinking mode=false2026.05 | 90.5 | 81.62 | 75.86 | |
| DeepSeek-V4-ProThinking mode=false2026.05 | 90.36 | 83.45 | 76.6 | |
| Hy-MT2-30B-A3BThinking mode=false2026.05 | 90.1 | 72.73 | 66.66 | |
| Gemma4-26B-A4BThinking mode=false2026.05 | 89.17 | 81.89 | 75.29 | |
| Gemma4-E4BThinking mode=false2026.05 | 86.81 | 78.04 | 71.73 | |
| Qwen3.6-35B-A3BThinking mode=false2026.05 | 84.84 | 78.12 | 71.48 | |
| Gemma4-E2BThinking mode=false2026.05 | 80.79 | 70.59 | 63.81 | |
| Hy-MT2-7BThinking mode=false2026.05 | 79.53 | 66.5 | 54.35 | |
| Hy-MT2-1.8BThinking mode=false2026.05 | 70.5 | 49.79 | 35.75 |