Why People Pay for DeepSeek Installation Packages (and How to Install It Yourself)

Word count 1353, about 7 minutes read

DeepSeek provides a family of open‑source large language models that can be downloaded and run locally.

Model lineup and specialties

DeepSeek-Coder – code completion, bug finding, and small program generation; 87% of training data is code.

DeepSeek-Math – solves high‑difficulty math problems with step‑by‑step explanations; performance comparable to GPT‑4 and Google Gemini.

DeepSeek-V3 – most capable model, supports reasoning, writing, data analysis; trained with a cost of over $5 million using more than 2 000 high‑end GPUs.

DeepSeek-R1 – cost‑effective series whose performance rivals commercial OpenAI models; available in sizes from 1.5 B to 671 B parameters.

Janus‑Pro‑7B – multimodal model for text‑to‑image, image‑to‑text, and image‑to‑image generation.

Choosing a model size

1.5B‑14B (mini) – runs on a phone, fast and low‑power; suitable for simple queries, short text generation, or weather checks.

32B‑70B (mid‑range) – behaves like a professional consultant; handles legal document analysis or industry‑report generation; requires a mid‑tier PC or server.

671B (giant) – “super‑student” capable of competition‑level problem solving, long‑form novel writing, and business decision analysis; requires top‑tier GPU servers.

Installation

Install Ollama

Download the installer for your operating system from the Ollama website, run it, and verify the installation: $ ollama --version Test the installation with a small model, e.g. deepseek-r1:1.5b:

$ ollama serve
$ ollama run deepseek-r1:1.5b  # (may take time to download in China)

After the service starts, the model can be accessed at http://localhost:11434 .

Deploy Open WebUI

Open WebUI is a self‑hosted AI platform that runs offline, supports multiple LLM runtimes (including Ollama), provides an OpenAI‑compatible API, and includes a built‑in RAG engine.

$ docker run -d -p 3000:8080 --network host -v open-webui:/app/backend/data \
  --name open-webui --restart always ghcr.io/open-webui/open-webui:main

After the container starts, open a browser and visit http://127.0.0.1:8080/ to use the UI.

References

[1] 5‑minute guide to building a local GitHub Copilot: https://mp.weixin.qq.com/s/cQPCeUEci6mjwFq3dEIwNg

[2] Ollama official site: https://ollama.com/

[3] Open WebUI plugin documentation: https://docs.openwebui.com/features/plugin/