List your company

Spreadsheet Understanding on SpreadsheetLLM benchmark (298 queries)

98AccuracyGPT-5
27.2845.646482.36Mar 6, 2026

Evaluation results

MethodLinks
GPT-5Retrieval Technique=BRTR2026.0398156,306273.36
Claude Sonnet 4.5Retrieval Technique=BRTR2026.0397166,117280.59
Claude Sonnet 4.5Retrieval Technique=SpreadsheetLLM2026.0391——
GPT-5Retrieval Technique=SpreadsheetLLM2026.0390——
Gemini 2.5 ProRetrieval Technique=SpreadsheetLLM2026.0389——
GPT-5Retrieval Technique=FRTR2026.0387——
Gemini 2.5 ProRetrieval Technique=FRTR2026.0385——
Claude Sonnet 4.5Retrieval Technique=FRTR2026.0384——
Gemini 2.5 ProRetrieval Technique=BRTR2026.037927,675223.79
GPT-4oRetrieval Technique=SpreadsheetLLM2026.0378——
GPT-4oRetrieval Technique=FRTR2026.0377——
LLaMA 4 Maverick 17BRetrieval Technique=SpreadsheetLLM2026.0373——
GPT-5Retrieval Technique=Naïve Full-Context2026.0368——
Gemini 2.5 ProRetrieval Technique=Naïve Full-Context2026.0367——
GPT-4oRetrieval Technique=BRTR2026.0356256,523315.93
Claude Sonnet 4.5Retrieval Technique=Naïve Full-Context2026.0355——
LLaMA 4 Maverick 17BRetrieval Technique=FRTR2026.0353——
LLaMA 4 Maverick 17BRetrieval Technique=Naïve Full-Context2026.0341——
GPT-4oRetrieval Technique=Naïve Full-Context2026.0335——
LLaMA 4 Maverick 17BRetrieval Technique=BRTR2026.033018,608408.73