OpenAI's most recent small reasoning model, providing high intelligence at the same cost and latency targets of o1-mini