DeepSeek-R1
Olympiad mathematics, logic verification, and autonomous thinking chains.
SWE-bench Verified
49.2%
MMLU-Pro
84%
MATH-500
97.3%
Context Window
128,000 tokens
| Specification | Value |
|---|---|
| Developer / Lab | DeepSeek-AI |
| Parameters Total | 671B |
| Parameters Active | 37B |
| Architecture Type | Pure RL Reasoning MoE + Cold-Start Thinking Tokens |
| License | MIT Permissive |
| Input Pricing (per 1M) | $0.55 |
| Output Pricing (per 1M) | $2.19 |