WifeBench is a humorous LLM ranking dashboard where the founder's wife scores language models 1-100 on 10 subjective questions she creates. It's a novelty product that plays on the absurdity of single-person subjective benchmarking against standardized metrics like MMLU.

One insight: Minimal traction (2 HN points, no corroborated signals), zero revenue model, and no defensibility make this a parody product unlikely to sustain demand or justify replication.