Subjective Evaluation of Research Ideas on 100 Research Ideas 10 Benchmark Topics
5.71Novelty ScoreEvoSci
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| EvoSciAgent Model=DeepSeek-v32026.05 | 5.71 | 4.68 | 5.25 | 5.15 | 4.9 | 3.95 | |
| VirSciAgent Model=DeepSeek-v32026.05 | 5.48 | 3.95 | 4.88 | 5.11 | 4.5 | 3.69 | |
| VirSciAgent Model=Qwen3-max2026.05 | 5.37 | 4.01 | 4.9 | 5 | 4.35 | 3.57 | |
| EvoSciAgent Model=Qwen3-max2026.05 | 5.14 | 4.98 | 5.2 | 4.89 | 4.72 | 3.81 | |
| VirSciAgent Model=GPT-4o2026.05 | 5.12 | 3.64 | 4.56 | 4.95 | 4.28 | 3.26 | |
| CoI-AgentAgent Model=DeepSeek-v32026.05 | 5.07 | 4.66 | 4.92 | 4.58 | 4.51 | 3.72 | |
| SciPIPAgent Model=Qwen3-max2026.05 | 4.87 | 5.41 | 4.89 | 4.53 | 4.54 | 3.17 | |
| EvoSciAgent Model=GPT-4o2026.05 | 4.78 | 4.62 | 5.01 | 4.75 | 4.45 | 3.44 | |
| SciPIPAgent Model=GPT-4o2026.05 | 4.72 | 4.76 | 4.74 | 4.49 | 4.26 | 3.06 | |
| CoI-AgentAgent Model=Qwen3-max2026.05 | 4.64 | 4.38 | 4.76 | 4.4 | 4.19 | 3.62 | |
| AI ScientistAgent Model=DeepSeek-v32026.05 | 4.6 | 6.85 | 5 | 4.29 | 4.68 | 3.39 | |
| CoI-AgentAgent Model=GPT-4o2026.05 | 4.52 | 4.77 | 4.79 | 4.33 | 4.2 | 3.21 | |
| SciPIPAgent Model=DeepSeek-v32026.05 | 4.42 | 5.8 | 4.85 | 4.13 | 4.34 | 3.02 | |
| AI ScientistAgent Model=GPT-4o2026.05 | 4.31 | 6.49 | 4.87 | 4.11 | 4.33 | 3.19 | |
| AI ScientistAgent Model=Qwen3-max2026.05 | 4.18 | 7 | 5.04 | 4.15 | 4.48 | 3.31 |