DeepSeek R1
deepseek-ai/DeepSeek-R1
Open Source · chat · open-weights
Context
163.8K
Max output
16K
Input $/1M
$0.70
Output $/1M
$2.50
Modalities
text
Released
20 Jan 2025
License: mit · deepseek-ai/DeepSeek-R1
AI summary
● machine-written
DeepSeek R1 open-weights reasoning model released January 20, 2025
DeepSeek R1 is an open-source reasoning model with 671B parameters (37B active) released under MIT license. It achieves performance comparable to OpenAI o1 while maintaining fully open reasoning tokens. The model is available as open weights for commercial use and distillation.
What's new
- 671B parameter model with 37B active parameters per inference pass
- Fully open reasoning tokens and weights under MIT license
- Open-source technical report published
- 163,840–164K token context window
- 16,000 token maximum output length
Best for
Complex reasoning and problem-solving tasksResearch and development requiring transparency into model reasoningCommercial applications with open-source licensing requirementsDeployments requiring full model control and customization