Surge AI Releases GDP.xlsx Benchmark; Top Agent Scores Just 38.3%

Surge AI has released GDP.xlsx, a follow-up to GDP.pdf, featuring 70 real-world spreadsheet tasks across 12 domains. The best-performing agent scored only 38.3%, highlighting significant room for improvement in handling complex Excel workbooks.

2026-10-01 ~ 2026-10-01 · 2 related posts

• GDP.xlsx benchmark: 70 real spreadsheet tasks, best frontier agent scores only 38.3% — echen · 2026-10-01
• Surge AI launches GDP.xlsx: 70 real-world tasks testing agents on messy spreadsheets — echen · 2026-10-01