Loading page content
DeepSeek-R1Open
Reasoning
Evaluation overview
Overall rank
#93
67.2 points · 180 models
Open-weight rank
#34
74 points · 109 models
Data coverage
37 items
3 sources · updated through 2026-07-16
Domain performance
Coding#75 / 160
70.4
Research#82 / 174
78.8
Agents#107 / 125
55.7
Evaluation highlights
All evaluationsBasic information
- Context (input / output)
- 164K / 66K
- Parameters
- 671B / 37B active
- Release date
- 2025-01
- License
- MIT
- Input modalities
- Text
- Output modalities
- Text
- Type
- Reasoning model
- Active parameters
- 37B
Availability
DeepSeekFull pricing
- Input
- ≈ $0.5714 /1M
- Output
- ≈ $2.2857 /1M
vLLM · FP8Deployment options
FP8 · 8 × 141 GB VRAM