Retrosynthesis Prediction on ORDerly in-distribution (test)
52.6Exact@1RetroReasoner
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| RetroReasonerTraining Paradigm=RL2026.03 | 52.6 | 82.6 | 72.4 | 95.2 | 78.6 | 3.186 | |
| RetroReasonerTraining Paradigm=SFT2026.03 | 51.2 | 81.2 | 73.4 | 94.4 | 76.5 | 3.898 | |
| Prediction-OnlyTraining Paradigm=RL2026.03 | 48.6 | 80.2 | 66.2 | 93.6 | 78.5 | 2.324 | |
| Prediction-OnlyTraining Paradigm=SFT2026.03 | 48.2 | 78.4 | 67.8 | 95 | 77.4 | 2.562 | |
| LlaSMolCategory=Molecular Prediction LLMs2026.03 | 25.4 | 66.8 | 58.4 | 92.6 | 43.7 | 7.814 | |
| Chem-RCategory=Molecular Reasoning LLMs2026.03 | 21.4 | 62.4 | 38.2 | 90.2 | 64.5 | 2.51 | |
| ChemDFMCategory=Molecular Reasoning LLMs2026.03 | 14 | 36 | 33.8 | 86.4 | 38 | 2.594 | |
| BioT5+Category=Molecular Prediction LLMs2026.03 | 9.8 | 38.8 | 16.6 | 66 | 36.6 | 1.198 | |
| ether0Category=Molecular Reasoning LLMs2026.03 | 3.4 | 48.8 | 12.8 | 75.8 | 48.7 | 1.33 | |
| Mol-LLMCategory=Molecular Prediction LLMs2026.03 | 1.6 | 3.6 | 6.6 | 15.4 | 1.5 | 0.192 | |
| GPT-oss-120BCategory=General Purpose LLMs2026.03 | 1.4 | 2.8 | 7 | 16.4 | 0.4 | 0.19 | |
| OpenAI-o3Category=General Purpose LLMs2026.03 | 1 | 2.2 | — | — | — | — | |
| GPT-5-miniCategory=General Purpose LLMs2026.03 | 1 | 3.8 | 15.2 | 49.8 | 3.1 | 0.794 | |
| Mol-InstructionsCategory=Molecular Prediction LLMs2026.03 | 0.8 | 3.6 | 3.8 | 18.8 | 0.7 | 0.29 | |
| PRESTOCategory=Molecular Prediction LLMs2026.03 | 0.6 | 18.6 | 1.2 | 45.8 | 43.4 | 0.496 | |
| Qwen3-235B-22BCategory=General Purpose LLMs2026.03 | 0.6 | 2.4 | 0.2 | 0.6 | 0 | 0.004 | |
| Qwen3-8BCategory=General Purpose LLMs2026.03 | 0 | 0.8 | 0.6 | 9.6 | 0.6 | 0.074 |