OpenAI inference cost reduction cut ChatGPT guest traffic from tens of thousands of Nvidia GPUs to just a couple hundred, ...
The change took some getting used to but now it's my workflow, not the GUI's ...