Question Answering on OpenBookQA (ACC, JCD, F&C, C&A, ARS)
98AccuracyGEMINI-3.1-PRO
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| GEMINI-3.1-PRO2026.05 | 98 | 86.33 | 89.95 | 84.42 | 88.63 | |
| DEEPSEEK-R12026.05 | 97.33 | 78.73 | 85.09 | 67.92 | 71.08 | |
| Self-Discover2026.05 | 96.67 | 89.47 | 90.98 | 88.21 | 90.41 | |
| MAVEN2026.05 | 96.33 | 92.92 | 93.77 | 89.92 | 96.1 | |
| DEEPSEEK-V3.22026.05 | 95.67 | 82.91 | 83.02 | 80.96 | 83.85 | |
| ReConcile2026.05 | 95.33 | 84.47 | 88.3 | 80.57 | 83.17 | |
| Self-Detection2026.05 | 95 | 86.6 | 89.85 | 86.34 | 89.98 | |
| Mixture-of-Agents2026.05 | 94.67 | 87.33 | 89.93 | 86.8 | 89.4 | |
| GPT-5.12026.05 | 94.33 | 83.22 | 87.45 | 79.32 | 84.35 | |
| LMVSLM2026.05 | 94 | 87.5 | 90.07 | 86.54 | 90.05 | |
| Multi-LLM Collab2026.05 | 94 | 88.7 | 90.51 | 87.87 | 90.49 |