Directory › Models › DeepSeek-R1 7B

DeepSeek-R1 7B

Models ✓ verified 🔒 local-only MIT
HuggingFace ↗ Website ↗ Ollama ↗

DeepSeek-R1 7B is a distilled reasoning model that provides step-by-step thinking, allowing users to run it locally.

Who it's for: This tool is suitable for individuals who want to utilize a self-hostable AI model for local processing.

Run it locally

ollama run deepseek-r1:7b

Strengths

  • Exposes step-by-step thinking
  • Runs locally
  • Available under the MIT license

Trade-offs

  • Limited availability of source code
  • No publicly available stars

Data locality

Runs fully offline — no data leaves your machine.

Hardware estimate

7.0B params · ~6.0 GB VRAM at Q4_K_M (estimate). Check your machine →