The R1-0528 model features enhanced reasoning capabilities, with DeepSeek asserting it’s closing the gap with OpenAI’s and Google’s latest models.