APEX-Agents · Investment Banking

World 246_MM_04

2/3Fail

APEX-Agents task World 246_MM_04 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 246Dual harnessGrader: rubric

Task IDtask_1f84a712cb2e4aaaa4b6778eeff49021

WorldInvestment Banking World 246

Expected outputmessage_in_console

Published runs4 models · dual config

Task prompt

What the agent was asked to do

Calculate the unlevered beta for Haleon (HLN) using Total Debt and Market Capitalization as of the end of FY2024. Assume 0.227 levered beta for HLN and a 21% Tax rate. Using the unlevered beta for HLN computed above, and the debt and equity values in the model, re-leverage the Beta for Kenvue and update the WACC with the new Re-levered Beta. Reply back with a message, giving the following results: - the New WACC - the New Implied Share Price. - the Variance in $ for Share Price (New-Original) Round all outputs to two decimal places.

Published trajectories

Agent runs on this task

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

Model	Harness	Score	Result	Links
GPT-5.5showcase	dual	2/3	Fail	Share page Public trace
Gemini 3.1 Pro	dual	0/3	Fail	Share page Public trace
GPT-5.4 mini	dual	1/3	Fail	Share page Public trace
GPT-5.4 nano	dual	0/3	Fail	Share page Public trace

Grading rubric

Criteria and grader verdict (showcase run)

States the New KVUE WACC is 5.08%
Pass
Evidence: <TEXT_RESPONSE> states “New WACC: 5.08%”. Assessment: Criterion “States the New KVUE WACC is 5.08%” is met exactly.
States New KVUE Implied Share Price is $34.72
Pass
Evidence: <TEXT_RESPONSE> states “New Implied Share Price: $34.72”. Assessment: Criterion “States New KVUE Implied Share Price is $34.72” is met exactly.
States the Variance in the New / Original share price is $17.23
Fail
Evidence: <TEXT_RESPONSE> states “Variance in Share Price (New - Original): +$17.24”. Assessment: Criterion requires the variance be stated as $17.23; the response states $17.24, so it does not match the required two-decimal value.