The open-source LLM reasoning benchmark for Performance Marketing.
Developed by hawky.ai — evaluating how well LLMs reason about real-world Meta Ads and Google Ads scenarios. v2 (494 questions) is built to surface the gap between knowledge recall and genuine reasoning.
PM-AGI v2 contains 494 expert-crafted questions across 4 categories and 5 reasoning types:
Meta Ads
227
Campaign structure… See the full description on the dataset page:
https://huggingface.co/datasets/Hawky-ai/pm-agi-benchmark.
