I’ve been running batches using the batch API and the cost seems to have drop significantly in January, althought from everything i can tell the workload has been the same. It went from 350$ average per month to now just 25$. Im using the gpt-4o-mini-2024-07-18 model. Did they announce a price drop or maybe there’s something im missing?
small update: I did some more research and it seems the token usage dropped significantly in january althought the batch loads are the same. Did openAI change how they count tokens? Im submitting images as well if that helps. Thanks!