Depending on the provider, GLM-5.2 is between 4.5-5x cheaper than Opus. You can compare prices/speed/etc. for basically all relevant models on aa https://artificialanalysis.ai/models/glm-5-2/providers

Hugging Face plus Z.ai API makes sense to me. Due to creators get paid, they can keep building better models, and the local-running community benefits from that over time.

AIhubmix currently is the cheapest rather than openrouter.

a company can just download GLM 5.2 and start self hosting this model using the chip designed and made by itself. That could lower the cost by 20-30x.

for hobbyist buying a few Mac Studio to host GLM 5.2 at home, the cost might 10x more than just using Opus API.