Before You Call an LLM Endpoint "Nerfed": A Small Statistics Checklist in Python

Every few weeks someone posts "provider X is serving a watered-down model" with a handful of...