Raycaster / evalsBack to SpreadsheetBench

SpreadsheetBench

55976

1/1Pass

SpreadsheetBench task 55976. Inspect the exact spreadsheet prompt, compare published model runs, open the agent response trace, and review grades for workbook editing.

Spreadsheet editingDual harness
ssb-55976
SpreadsheetBench
2 models
Dual (parsed + original)
PromptWorkbook objective and answer cells
ResponseOpen each trace to inspect tool use and edits
GradeCompare score and pass/fail by model

Task prompt

What the agent was asked to do

You are solving a spreadsheet benchmark task in a real workbook. Objective: Produce the correct final workbook state for the expected answer region. What matters: - Only the values in the expected answer region will be graded. - The workbook is the answer. Instructions: 1. Read the workbook and inspect the relevant data region first. 2. Infer the required result for the provided workbook instance. 3. Write the final value(s) directly into the expected answer region. 4. Do not rely on prose, formulas in your chat response, pseudocode, or VBA as the answer unless the benchmark explicitly requires those to be written into cells. 5. If the natural-language task asks for a general method, formula, or macro, convert that into the concrete result needed for this workbook instance. 6. Keep your final text response short and only summarize the workbook cells you changed. Relevant data region(s): Expected answer region(s): F25:I25 Expected answer sheet(s): Task: Tally the 'Yes' responses from an 80-question survey in Excel, where each question requires users to mark an 'x' under 'Yes' if they agree, and associate each 'Yes' response with one of four columns that categorize respondents. Each column totals the number of 'x' marks for its set of associated questions and summarizes them in a cell (F25 for the first column, G25 for the second, and so on). For instance, if a respondent marks 'x' under 'Yes' for questions 2, 4, 6, 10, 17, etc., each 'x' would count as a point in column F, with the total displayed in F25. The goals are similar for the other columns but with different sets of questions (like 7, 13, 15, 16, etc. for column G summarized in G25). Build formulas or a method I could use in Excel to achieve this summary. Make sure to include row 4 values to count. Do not alter any formatting when making changes.

Published trajectories

Agent runs on this task

Curated dual-harness runs. Best scored run per model.

ModelHarnessScoreResultLinks
GPT-5.4showcasedual1/1Pass
Gemini 3.1 Produal1/1Pass