Overview
Advanced reasoning model matching OpenAI o1 performance in logical, mathematical, and coding challenges with transparent thinking process and cost efficiency
Key Features
- First-generation reasoning model trained via large-scale reinforcement learning (RL)
- Competitive with OpenAI o1 in math, coding, and reasoning at 95% lower cost
- Self-verification, reflection, and chain-of-thought (CoT) reasoning capabilities
- Includes 6 distilled models (1.5B-70B) for resource-constrained environments
- MIT licensed with full commercial use, modification, and distillation rights
Input Capabilities
Text
Technical Specifications
Context Window
64,000
Max Output
64,000
Model Information
Provider: DeepSeek
Model Code: deepseek-reasoner
Category: Chat