Build a Multi-Model Evaluation CLI in Node.js

Testing a prompt against one model in a chat interface tells me almost nothing about how it will behave in production. The output may look good, but I still do not know: whether another model follows