Spreadsheet Reasoning on SpreadsheetBench
50.5Pass@1SkillTTA
Evaluation Results
| Method | Links | |
|---|---|---|
| SkillTTASkill Synthesis Model=GPT-4o2026.05 | 50.5 | |
| SkillTTASkill Synthesis Model=GPT-4o-mini2026.05 | 44.3 | |
| Trace2SkillSkill Synthesis Model=GPT-4o-mini2026.05 | 42 | |
| ReActSkill Synthesis Model=None2026.05 | 41.8 | |
| Trace2SkillSkill Synthesis Model=GPT-4o2026.05 | 39.7 | |
| Zero-shotSkill Synthesis Model=None2026.05 | 36.3 | |
| MemRLSkill Synthesis Model=None2026.05 | 28.7 |