# 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.



## Model Information

- **Organization**: [DeepSeek](/llm.txt)
- **Slug**: deepseek-r1-zero
- **Available at Providers**: 2
- **Release Date**: January 20, 2025

### Benchmark Scores
- GPQA: 0.733

## Providers

| Provider | Name | $ Input (per 1M) | $ Output (per 1M) | Free | Link |
|----------|------|-----------------|------------------|------|------|
| [AIHubMix](/llm/aihubmix.txt) | DeepSeek-R1-Zero | 2.20 | 2.20 |  | [View](https://aihubmix.com/model/deepseek-ai/DeepSeek-R1-Zero) |
| [Writingmate](/llm/writingmate.txt) | DeepSeek: DeepSeek R1 Zero |  |  |  | [View](https://writingmate.ai/models/deepseek/deepseek-r1-zero) |

---

[← Back to all providers](/llm.txt)