SpreadsheetBench · GPT-5.4 · dual
44017
GPT-5.4 on SpreadsheetBench: 44017 (dual harness). Browse score, rubric, and public trace.
Prompt excerpt
Task context
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): AD14:AO42 Expected answer sheet(s): Task: I need to calculate updated rates for each customer based on price increases that roll out over time. Each customer has baseline rates in a 12-month historical range, an effective start date when increases begin applying, and a frequency value (representing months between waves) that determines when each increase takes effect. There are up to four sequential price increase percentages defined in separate columns.Create formulas in columns AD through AO starting at row 14 that calculate new rates by applying cumulative price increases to the base rate in column W. The increases should phase in based on the effective date in column L, the frequency in column J, and the increase percentages in columns M through P. Each column represents a different month with dates in row 9. Only show values when the column date is on or after the effective date. There should be no yellow fill in the cells you complete.
Response trace
Agent response, tools, files, and edits
On mobile, open the full trace for the best view of the agent response, tool calls, file reads, edits, and grading context.
Open mobile trace