When is Serverless Inference Cheaper than Your Self Hosted GPU? I Benchmarked gpt-oss-120b on Both

If you run LLM inference in production, you eventually will ask yourself, should you rent a GPU and run the model yourself, or do you use a serverless API and pay per token? Everyone has an opinion. Far fewer people show you the actual numbers that decide it. So I ran both. I put the same model, gpt