Raycaster / evalsBack to AI Agents for Tax Due Diligence

APEX-Agents · Investment Banking

World226_BS_01

5/6Fail

APEX-Agents task World226_BS_01 in AI Agents for Tax Due Diligence. Compare dual-harness agent runs across models — rubric criteria, scores, and public traces.

AI Agents for Tax Due DiligenceInvestment Banking World 226Dual harnessGrader: rubric
Task IDtask_ae92292a5cfb4594ae8746ee31ce498b
WorldInvestment Banking World 226
Expected outputmessage_in_console
Published runs4 models · dual config

Task prompt

What the agent was asked to do

Planet Fitness is looking to divest its entire 281 stores, which it owns as of September 30, 2025, to a franchise owner. Round all results to two decimal places and present it in $mm. Using the LBO model, perform a DCF analysis for the company as per the base case scenario for the projected cash flows for the 281 stores, and assume the following: 1) Assume that the average revenue per store increases by 5% YoY for every quarter from Q4 2025 through the end of 2030 2) Assume EBITDA margin for the business remains at 39% every quarter from Q4 2025 through the end of 2030 2) Assume that the effective tax rate is 20% 3) Assume that the depreciation rate is 5% of the revenue 4) Assume that maintenance capex is 2% of sales and there is no growth capex 5) Assume a discount rate of 12% and terminal growth rate of 2% 6) Do the enterprise valuation as of December 31, 2025 Print here the FCFF for 2026 to 2030. Also give the Enterprise Value of the corporate-owned store business

Published trajectories

Agent runs on this task

Curated dual-harness runs (parsed + original sandbox). Best scored run per model.

ModelHarnessScoreResultLinks
GPT-5.5showcasedual5/6FailShare pagePublic trace
Gemini 3.1 Produal5/6FailShare pagePublic trace
GPT-5.4 minidual5/6FailShare pagePublic trace
GPT-5.4 nanodual0/6FailShare pagePublic trace

Grading rubric

Criteria and grader verdict (showcase run)

  1. States 2026 FCFF is $173.68 mm

    Pass

    Evidence: TEXT_RESPONSE table lists “2026 | $173.68”. Assessment: Criterion requires stating 2026 FCFF is $173.68 mm; pass because the response states exactly that value in $mm.

  2. States 2027 FCFF is $182.37 mm

    Pass

    Evidence: TEXT_RESPONSE table lists “2027 | $182.37”. Assessment: Criterion requires stating 2027 FCFF is $182.37 mm; pass because the response states exactly that value in $mm.

  3. States 2028 FCFF is $191.48 mm

    Pass

    Evidence: TEXT_RESPONSE table lists “2028 | $191.48”. Assessment: Criterion requires stating 2028 FCFF is $191.48 mm; pass because the response states exactly that value in $mm.

  4. States 2029 FCFF is $201.06 mm

    Pass

    Evidence: TEXT_RESPONSE table lists “2029 | $201.06”. Assessment: Criterion requires stating 2029 FCFF is $201.06 mm; pass because the response states exactly that value in $mm.

  5. States 2030 FCFF is $211.11 mm

    Pass

    Evidence: TEXT_RESPONSE table lists “2030 | $211.11”. Assessment: Criterion requires stating 2030 FCFF is $211.11 mm; pass because the response states exactly that value in $mm.

  6. States Enterprise Value of the corporate-owned store business is $1,775.26 mm

    Fail

    Evidence: TEXT_RESPONSE states “Enterprise Value of the 281 corporate-owned store business: $1,906.17mm.” Assessment: Criterion requires stating Enterprise Value is $1,775.26 mm; fail because the response gives $1,906.17 mm, not the required value.