DeepSeek R1
DeepSeekยทdeepseek/deepseek-r1
Frontier open-weights reasoning model utilizing large-scale reinforcement learning without supervised fine-tuning cold start.
Open WeightsOpenReasoningTool CallingStructured OutputsNatural Language ProcessingArtificial Intelligence
Context
131,072
Output Limit
65,536
Linked Papers
42
Citations
1.4K
Release Date
Jan 20, 2025
Model Specifications
- Lab / Vendor
- DeepSeek
- Model Family
- DeepSeek
- Modality
- text
- Architecture
- Mixture-of-Experts (MoE) + Multi-Head Latent Attention
- Parameters
- 671B MoE (37B active)
- Context Window
- 131,072
- Max Output
- 65,536
- Input / Output Price
- $0.55 / $2.19 per 1M
- License
- MIT
Research & Evaluation Papers
Academic evaluations, empirical benchmarks, and publications for DeepSeek R1.
Sort