Spreadsheet Question Answering on FRTR-Bench full 1.0
74AccuracyFRTR
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| FRTRBackbone=Claude Sonnet 4.52026.01 | 74 | 11.71 | |
| FRTRBackbone=GPT-52026.01 | 73 | 15.5 | |
| FRTRBackbone=Gemini 2.5 Pro2026.01 | 67 | 26.9 | |
| FRTRBackbone=LLaMA Maverick-17B2026.01 | 56 | 2.42 | |
| FRTRBackbone=GPT-OSS-120B2026.01 | 51 | 5.75 | |
| FRTRBackbone=GPT-4o2026.01 | 49 | 5.04 | |
| SpreadsheetLLMBackbone=Gemini 2.5 Pro2026.01 | 24 | 33.35 | |
| SpreadsheetLLMBackbone=LLaMA Maverick-17B2026.01 | 23 | 1.15 | |
| SpreadsheetLLMBackbone=GPT-OSS-120B2026.01 | 21 | 8.8 | |
| SpreadsheetLLMBackbone=GPT-52026.01 | 18 | 18.1 | |
| SpreadsheetLLMBackbone=Claude Sonnet 4.52026.01 | 13 | 8.8 | |
| SpreadsheetLLMBackbone=GPT-4o2026.01 | 6 | 1.14 |