Download: v1.0.0v1.0.0 ↓1.0.02025-06-27
+9
Daily
+26
Weekly
92,029
stars
11,690
forks
598
watchers
60
prs
740
issues
1
releases

README

DeepSeek-R1





  



  
  
  
  
  
  
  
  
  
  
  Paper Link👁️


 1. Introduction

We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1. 
DeepSeek-R1-Zero, a model trained via large-scale reinforcement learning (RL) without supervised fine-tuning (SFT) as a preliminary step, demonstrated remarkable performance on reasoning.
With RL, DeepSeek-R1-Zero naturally emerged with numerous powerful and interesting reasoning behaviors.
However, DeepSeek-R1-Zero encounters challenges such as endless repetition, poor readability, and language mixing. To address these issues and further enhance reasoning performance,
we introduce DeepSeek-R1, which incorporates cold-start data before RL.
DeepSeek-R1 achieves performance comparable to OpenAI-o1 across math, code, and reasoning tasks. 
To support the research community, we have open-sourced DeepSeek-R1-Zero, DeepSeek-R1, and six dense models distilled from DeepSeek-R1 based on Llama and Qwen. DeepSeek-R1-Distill-Qwen-32B outperforms OpenAI-o1-mini across various benchmarks, achieving new state-of-the-art results for dense models.

NOTE: Before running DeepSeek-R1 series models locally, we kindly recommend reviewing the Usage Recommendation section.


  


 2. Model Summary

---

Post-Training: Large-Scale Reinforcement Learning on the Base Model

-  We directly apply reinforcement learning (RL) to the base model without relying on supervised fine-tuning (SFT) as a preliminary step. This approach allows the model to explore chain-of-thought (CoT) for solving complex problems, resulting in the development of DeepSeek-R1-Zero. DeepSeek-R1-Zero demonstrates capabilities such as self-verification, reflection, and generating long CoTs, marking a significant milestone for the research community. Notably, it is the first open research to validate that reasoning capabilities of LLMs can be incentivized purely through RL, without the need for SFT. This breakthrough paves the way for future advancements in this area.

-   We introduce our pipeline to develop DeepSeek-R1. The pipeline incorporates two RL stages aimed at discovering improved reasoning patterns and aligning with human preferences, as well as two SFT stages that serve as the seed for the model's reasoning and non-reasoning capabilities.
    We believe the pipeline will benefit the industry by creating better models. 

---

Distillation: Smaller Models Can Be Powerful Too

-  We demonstrate that the reasoning patterns of larger models can be distilled into smaller models, resulting in better performance compared to the reasoning patterns discovered through RL on small models. The open source DeepSeek-R1, as well as its API, will benefit the research community to distill better smaller models in the future. 
- Using the reasoning data generated by DeepSeek-R1, we fine-tuned several dense models that are widely used in the research community. The evaluation results demonstrate that the distilled smaller dense models perform e

Top issues

Latest release

v1.0.0 v1.0.0 2025-06-27
This release is created for archival purposes and DOI generation.

News & coverage (22)

Hacker NewsDeepSeek-R1▲ 1,843 · 663 comments · 2025-01-20Hacker NewsWhy is the GitHub trending page weirdly excluding DeepSeek projects?▲ 9 · 1 comments · 2026-08-16Hacker NewsDeepseek R1 Zero learns to reason using reinforcement learning on base model [pdf]▲ 6 · 0 comments · 2025-01-20Hacker NewsDeepseek R1 paper – at 2% of the price of OpenAI o1 [pdf]▲ 5 · 0 comments · 2025-01-23Hacker NewsDeepSeek drops recommended R1 deployment settings▲ 2 · 0 comments · 2025-02-14Hacker NewsDeepseek open models seem to follow CCP party-line▲ 2 · 0 comments · 2025-01-27eu.36krThe paper of DeepSeek-R1 has made it onto the cover of Nature. The corresponding author is LIANG Wenfeng.2025-09-17githubDeepSeek-R1-0528 is now generally available in GitHub Models2025-06-04VentureBeatDeepSeek R1-0528 arrives in powerful open source challenge to OpenAI o3 and Google Gemini 2.5 Pro2025-05-29CNETDeepSeek R1 First Impressions: I’ve Seen How the Machine Thinks2025-02-09
12 moredevblogsBuild Intelligent Apps with .NET and DeepSeek R1 Today!2025-01-31theregisterMicrosoft catapults DeepSeek R1 into Azure AI Foundry, GitHub2025-01-30cybersecuritynewsDeepSeek is Now Available With Microsoft Azure AI Foundry & GitHub2025-01-30Searchenginejournal.comDeepSeek-R1: The Open-Source AI Challenging ChatGPT2025-01-30githubDeepSeek-R1 is now available in GitHub Models (Public Preview)2025-01-29techtargetMicrosoft, AWS and Cerebras launch DeepSeek-R1 model2025-01-29The VergeMicrosoft makes DeepSeek’s R1 model available on Azure AI and GitHub2025-01-29Ars TechnicaHow does DeepSeek R1 really fare against OpenAI’s best reasoning models?2025-01-28BGRDevelopers Caught DeepSeek R1 Having An 'Aha Moment' On Its Own During Training2025-01-27METRDetails about METR's preliminary evaluation of DeepSeek-R12025-03-05TechRadarDeepSeek R1 is now available on Nvidia, AWS, and Github as available models on Hugging Face shoot past 3,0002025-02-03Microsoft AzureDeepSeek R1 is now available on Azure AI Foundry and GitHub2025-01-29

Snapshot: 2026-11-16T09:23:14Z · Data license Apache-2.0 · JSON