Raycaster/ Eval

APEX-Agents · Management Consulting

World126_JD_04

Best published1/4Fail

APEX-Agents task World126_JD_04 in AI Agents for ESG and Climate Risk Analysis. Compare dual-harness agent runs across models, scores, and public traces.

AI Agents for ESG and Climate Risk AnalysisManagement Consulting World 113.1Dual harnessGrader: rubric
task_c5080c2d60fa457faeb309841b8b442a
Management Consulting World 113.1
message_in_console
6 models · dual config

Task prompt

What the agent was asked to do

Please check how KO and MDLZ differ in expected upside once we apply the ESG and GLP-1 filters and account for each investor’s maximum allowable ESG risk level. Use the survey data and the ESG risk thresholds to determine which respondents are eligible to hold each company. Then calculate the confidence weighted average and standard deviation of expected annual return for KO and MDLZ. Show each company’s weighted average and standard deviation of expected annual returns. Round only the final results, going to two decimal places.

Published trajectories

Agent runs on this task

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

ModelHarnessScoreResultLinks
GPT-5.4 minidual1/4Fail
Gemini 3 Flashdual0/4Fail
Gemini 3.1 Produal0/4Fail
GPT-5.4dual0/4Fail
GPT-5.4 nanodual0/4Fail
GPT-5.5dual0/4Fail

Grading rubric

Rubric criteria

Runs are graded against these criteria. Open a run for model-specific verdicts.

  1. States that the Weighted Avg Expected Annual Return for KO is 6.31

  2. States that the Weighted Avg Expected Annual Return for MDLZ is 6.71

  3. States that the Weighted Stdev Expected Annual Return for KO is 2.59

  4. States that the Weighted Stdev Expected Annual Return for MDLZ is 2.77