Molecule Editing on ChemCoTBench
100Addition Success RateGemini-2.5-Pro-think
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Gemini-2.5-Pro-think2026.01 | 100 | 85 | 81.7 | |
| Claude3.7-sonnet-think2026.01 | 85 | 80 | 83.4 | |
| ChemCRAFT2026.01 | 85 | 95 | 80 | |
| DeepSeek-R12026.01 | 70 | 70 | 68.3 | |
| Deepseek-V32026.01 | 70 | 75 | 76.7 | |
| O3-Mini2026.01 | 65 | 55 | 80 | |
| Qwen2.5-32B-Instruct2026.01 | 50 | 50 | 48.3 | |
| Qwen3-235B-A22B2026.01 | 40 | 75 | 66.7 |