DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrates remarkable reasoning capabilities. Through RL, DeepSeek-R1-Zero naturally emerges with numerous powerful and intriguing reasoning behaviors.

DeepSeek topped the Apple App Store chart and sparked fears the Chinese company was quickly catching up with OpenAI's ChatGPT while costing far less.