I don’t know what’s going on with those GPT 6.2 prices. There should be some standardization auditable in the x-axis cost per task. If not, labs may game that through loss leadership tactics.