DeepSeek: DeepSeek R1 Zero

DeepSeek-R1-Zero is a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step. It's 671B parameters in size, with 37B active in an inference pass. It demonstrates remarkable performance on reasoning. With RL, DeepSeek-R1-Zero naturally emerged with numerous powerful and interesting reasoning behaviors. DeepSeek-R1-Zero encounters challenges such as endless repetition, poor readability, and language mixing. See [DeepSeek R1](/deepseek/deepseek-r1) for the SFT model.

• Context window: 163,840 tokens
• Input: text
• Output: text

View on OpenRouter. Model data sourced from OpenRouter.