GPT-4o-mini isn't cheap. This model costs half as much.

Most devs default to GPT-4o-mini as the "cheap" option. I ran the real blended math (60% output tokens) — the actual winner is Mistral Small 4 at $0.22/M. That's half the price. And almost nobody uses it.

Data source: artificialanalysis.ai/models

Видео GPT-4o-mini isn't cheap. This model costs half as much. канала Hype Check