| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| Scientific Ideation 60 samples human-validated (test) | Novelty3.63 | 9 | 4mo ago | ||
| 30 open research questions from PhD candidates (test) | Our Method | Average Novelty3.43 | 7 | 3mo ago | |
| NeurIPS Overall 2025 | Agentic-Ideation | Novelty6.74 | 6 | 24d ago | |
| NeurIPS 2025 (Others) | Agentic-Ideation | Novelty6.92 | 6 | 24d ago | |
| NeurIPS 2025 (CV) | Agentic-Ideation | Novelty6.74 | 6 | 24d ago | |
| Scientific Ideation (Human Evaluation) | Agentic-Ideation | Novelty6.71 | 6 | 24d ago | |
| Scientific Ideation Out-of-Domain | SciThinker-30B | GPT-5.2 Score59 | 2 | 4mo ago | |
| scientific papers Out-of-Domain | SciThinker-30B | Win Rate vs GPT-5.259 | 2 | 4mo ago | |
| scientific papers In-Domain | SciThinker-30B | Win Rate vs GPT-5.261 | 2 | 4mo ago |
