MerchBench is an open-source benchmark for retail AI model routing. It helps retailers evaluate which model or workflow tier is economically sufficient for different classes of merchandise-planning decisions.
The core idea is simple: retail AI should not be governed by model leaderboards alone. It should be governed by decision risk, reversibility, economic impact, deterministic controls, and human-review policy.
If you are searching for retail AI evaluation, LLM… See the full description on the dataset page:
https://huggingface.co/datasets/Novice-ninja/retail-merchbench.
