Tutoring Evaluation on Tutoring Evaluation (test)
95.4AccuracyGemini-3-Flash
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Gemini-3-FlashAccess=Closed-source2026.05 | 95.4 | 93.9 | 97.9 | 97.7 | 96.4 | 74.8 | 97.5 | 97.8 | 93.9 | |
| PEARL-30B (Ours)Size=30B2026.05 | 94 | 89.3 | 96.1 | 94.4 | 88.6 | 95 | 90.7 | 95.1 | 92.9 | |
| GPT-5.1Access=Closed-source2026.05 | 93.4 | 84.9 | 94.8 | 94.6 | 91.1 | 89 | 92 | 94.6 | 91.8 | |
| DeepSeek-V3.2Access=Open-source2026.05 | 89.8 | 76.6 | 93.7 | 91.6 | 87.5 | 57.4 | 87.5 | 93.1 | 84.7 | |
| Qwen3-235B-A22BAccess=Open-source, Size=235B2026.05 | 89.6 | 56.3 | 95.4 | 90.2 | 75.4 | 49.9 | 81.4 | 95 | 79.2 | |
| MiniMax-M2.1Access=Open-source2026.05 | 89.5 | 59.5 | 94 | 92.4 | 81.1 | 58 | 83.8 | 93.8 | 81.5 | |
| Qwen3-30B-A3B (Baseline)Size=30B, Architecture=A3B2026.05 | 87.9 | 63.8 | 93.1 | 89.3 | 79.1 | 50 | 81.2 | 92.6 | 79.6 | |
| Kimi-K2-InstructAccess=Open-source2026.05 | 83.4 | 66.7 | 87.7 | 87 | 77.6 | 43.1 | 79.2 | 86.5 | 76.4 | |
| GLM-4.7Access=Open-source2026.05 | 82.7 | 84.3 | 86.2 | 85.6 | 81.3 | 52.4 | 82.6 | 86.5 | 80.2 | |
| MuduoLLM-14BAccess=Open-source, Size=14B2026.05 | 59 | 46.2 | 73.5 | 66.6 | 54.5 | 42.6 | 57.2 | 73.8 | 59.2 | |
| InnoSpark-plus-72BAccess=Open-source, Size=72B2026.05 | 56.4 | 37 | 67.9 | 57.6 | 40.7 | 40.9 | 45.7 | 68.8 | 51.9 | |
| InnoSpark-turbo-7BAccess=Open-source, Size=7B2026.05 | 41.7 | 29.9 | 65.6 | 51.7 | 40.9 | 39.4 | 42.2 | 66.4 | 47.2 |