Yeah it’s really annoying how opaque they are about the model there. Always just ”GPT 4 based” or ”GPT 4o based” if you dig in their blog archives. Makes one un

Yeah it’s really annoying how opaque they are about the model there. Always just ”GPT 4 based” or ”GPT 4o based” if you dig in their blog archives. Makes one unable to check it against benchmarks or see when it’s updated. Setting expectations. Is it a distill? Lower precision quant? An old revision? Who knows.