DeepSeek claims that its R1 model performs at par with leading models like OpenAI's GPT and excels in logical reasoning tasks.