Idea Generation Evaluation on SGI-bench
80.53NoveltyDeepInnovator
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| DeepInnovatorOpponent Model=Qwen-14B-Instruct2026.02 | 80.53 | 87.61 | 92.92 | 93.81 | |
| DeepInnovatorOpponent Model=Minimax-M2.12026.02 | 59.29 | 10.62 | 50.44 | 61.06 | |
| DeepInnovatorOpponent Model=GPT-4o2026.02 | 49.56 | 11.5 | 76.11 | 44.25 | |
| DeepInnovatorOpponent Model=Deepseek-r12026.02 | 47.79 | 23.01 | 45.13 | 59.29 | |
| DeepInnovatorOpponent Model=Grok-4.12026.02 | 37.17 | 10.62 | 43.36 | 45.13 | |
| DeepInnovatorOpponent Model=Qwen3-max2026.02 | 32.74 | 19.47 | 66.37 | 91.15 | |
| DeepInnovatorOpponent Model=Gemini-2.5-pro2026.02 | 15.93 | 18.58 | 43.36 | 68.14 |