All AutosheetBench tasks

Clear all formatting, keep the data

FormattingEasyLLM judge scoringLatest run

Passed by all 12 products on their best model.

Prompt and starting spreadsheet

Task files

Prompt

Clear all the formatting on this sheet.

Starting spreadsheet

Sheet1
ABCDE
1RepRegionDealsRevenueTarget
2AliceWest24$148,000$120,000
3BobEast12$82,000$90,000
4CarlaWest31$205,000$160,000
5DanNorth9$61,000$75,000
6ErinSouth27$173,000$140,000
7FrankEast18$99,000$100,000

Expected result

How each AI configuration did

21 of 22 passRun files
Excel and Sheets · All types · All models · All outputs

Pass criteria

All must be met, checked by an LLM judge.
  • All cell formatting is removed
  • Both conditional formatting rules are removed
  • The data is unchanged

Passed 21

Autosheet in Excel Opus 5.5 · Medium

Passed7s · 2¢

Autosheet in Excel GPT-6 Astra · Low

Passed5s · 3¢

Autosheet in Sheets Opus 5.5 · Medium

Passed12s · 3¢

Autosheet in Excel Sonnet 5.5 · Medium

Passed3s · 0.8¢

Claude in Excel Opus 5.5 · High

Passed24s · 19¢

Autosheet in Excel Haiku 5.5 · High

Passed5s · 0.1¢

Claude in Sheets Opus 5.5 · Medium

Passed24s · 6¢

Claude Code + Sheets MCP Opus 5.5 · Medium

Passed41s · N/A

Autosheet in Excel GPT-6.1 Sol · Medium

Passed5s · 0.9¢

Claude Code + Autosheet MCP Opus 5.5 · Medium

Passed27s · N/A

ChatGPT + Autosheet MCP GPT-6 · Medium

Passed40s · N/A

Autosheet in Sheets Haiku 5.5 · High

Passed23s · 0.2¢

ChatGPT in Excel GPT-6 Astra · Medium

Passed9s · 36¢

Claude in Excel Haiku 5.5 · High

Passed15s · 0.5¢

Claude in Sheets Haiku 5.5 · High

Passed33s · 2¢

Autosheet in Excel GPT-6 Luna · High

Passed5s · 0.1¢

Copilot in Excel Auto

Passed17s · N/A

ChatGPT + Drive MCP GPT-6 · Medium

Passed28s · N/A

ChatGPT in Excel GPT-6 Luna · High

Passed15s · 0.3¢

Gemini in Sheets Auto

Passed1m 18s · N/A

ChatGPT in Sheets GPT-6 Astra · Medium

Passed27s · 34¢

Failed 1

ChatGPT in Sheets GPT-6 Luna · High

Failed9s · 0.3¢

Why it failed: Number formats were not reset to default General format; Revenue and Target still render as currency ($148,000, etc.) rather than plain numbers.