All AutosheetBench tasks

Make a donut chart of spend

ChartsHardLLM judge scoringLatest run

Passed by 8 of 12 products on their best model. Failed: Claude Code + Sheets MCP, Claude Code + Autosheet MCP, ChatGPT + Drive MCP and ChatGPT in Sheets.

Prompt and starting spreadsheet

Task files

Prompt

Make a donut chart of spend by category.

Starting spreadsheet

Spend
AB
1CategorySpend
2People420000
3Software165000
4Marketing118000
5Facilities92000
6Travel54000

Expected result

How each AI configuration did

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

Pass criteria

All must be met, checked by an LLM judge.
  • A donut chart of spend by category
  • One correctly sized slice per category
  • A readable title and labels

Passed 15

Autosheet in Excel Opus 5.5 · Medium

Passed9s · 4¢

Autosheet in Excel GPT-6 Astra · Low

Passed12s · 7¢

Autosheet in Sheets Opus 5.5 · Medium

Passed14s · 5¢

Autosheet in Excel Sonnet 5.5 · Medium

Passed6s · 2¢

Claude in Excel Opus 5.5 · High

Passed14s · 9¢

Autosheet in Excel Haiku 5.5 · High

Passed9s · 0.2¢

Claude in Sheets Opus 5.5 · Medium

Passed16s · 4¢

Autosheet in Excel GPT-6.1 Sol · Medium

Passed15s · 3¢

ChatGPT + Autosheet MCP GPT-6 · Medium

Passed32s · N/A

ChatGPT in Excel GPT-6 Astra · Medium

Passed13s · 36¢

Claude in Excel Haiku 5.5 · High

Passed11s · 0.5¢

Claude in Sheets Haiku 5.5 · High

Passed23s · 1¢

Autosheet in Excel GPT-6 Luna · High

Passed13s · 0.3¢

Copilot in Excel Auto

Passed1m 2s · N/A

Gemini in Sheets Auto

Passed27s · N/A

Failed 7

Claude Code + Sheets MCP Opus 5.5 · Medium

Failed27s · N/A

Why it failed: The chart lacks data labels showing spend amounts or percentages.

Claude Code + Autosheet MCP Opus 5.5 · Medium

Failed52s · N/A

Why it failed: Chart lacks data labels showing percentages or spend amounts.

Autosheet in Sheets Haiku 5.5 · High

Failed22s · 0.2¢

Why it failed: Missing data labels showing values or percentages.

ChatGPT + Drive MCP GPT-6 · Medium

Failed26s · N/A

Why it failed: No data labels displaying percentage or currency amounts are included on the chart slices.

ChatGPT in Excel GPT-6 Luna · High

Failed35s · 0.6¢

Why it failed: Chart lacks data labels showing percentage or spend amounts.

ChatGPT in Sheets GPT-6 Astra · Medium

Failed22s · 31¢

Why it failed: No doughnut chart was created.

ChatGPT in Sheets GPT-6 Luna · High

Failed34s · 0.4¢

Why it failed: No doughnut chart was created in the workbook.