I don’t know why people think DeepSeek did reasoning models / RLVR before OpenAI, there was a gap of months.