Introduction

In December of last year, DeepSeek’s DeepSeek-V3 caused a huge stir in the global AI field. With extremely low training costs, it achieved performance comparable to top-tier models like GPT-4 and Claude Sonnet 3.5, stunning the industry. Tencent Tech had conducted an in-depth analysis of this model, explaining its technical background for achieving low-cost and high-efficiency results using simple and straightforward language.

This time is different; the newly released DeepSeek-R1 not only maintains the low cost but also sees significant advancements in technology. Moreover, it is an open-source model. This new model continues to leverage its high cost-effectiveness, achieving GPT-4-level performance at just a tenth of the cost. Consequently, many professionals have even raised the slogan that "DeepSeek will replace OpenAI."

This tutorial includes:

1. Installing ollama
2. Downloading and using DeepSeek models in ollama
3. Interacting with Chatbox for chat functionality

Code exmaple

Installing ollama

Require sudo

sudo curl -fsSL https://ollama.com/install.sh | sh

Downloading and using DeepSeek models in ollama

1. Start ollama

   ollama start

   or start and put it in the background

   ollama start&

2. Choose one version to download and run
   DeepSeek-R1-Distill-Qwen-1.5B
   Suggest 4-8GB graphics card memory or a multi-threaded CPU

   ollama run deepseek-r1:1.5b

   DeepSeek-R1-Distill-Qwen-7B
   Suggest 8GB graphics card memory

   ollama run deepseek-r1:7b

   DeepSeek-R1-Distill-Llama-8B
   Suggest 8GB graphics card memory

   ollama run deepseek-r1:8b

   DeepSeek-R1-Distill-Qwen-14B
   Suggest 14GB graphics card memory

   ollama run deepseek-r1:14b

   DeepSeek-R1-Distill-Qwen-32B
   Suggest 24GB graphics card memory

   ollama run deepseek-r1:32b

   DeepSeek-R1-Distill-Llama-70B
   Suggest 4x RTX 4090 or 2x A100 total 80GB

   ollama run deepseek-r1:70b

   DeepSeek-R1 model
   Suggest 8x A100/H100

   ollama run deepseek-r1:671b

   Path of model storage in linux: ~/.ollama/models/
   Start a Conversation
   

3. Stop ollama

   pkill -f ollama

Interacting with Chatbox for chat functionality

1. Download app from website
   https://chatboxai.app/zh#download
   choose linux 64bit version to download
   or copy link to download

   wget https://download.chatboxai.app/releases/Chatbox-1.9.8-x86_64.AppImage

2. Give software running permissions. File with AppImage format could be run directly.

   chmod a+x Chatbox-1.9.8-x86_64.AppImage

   Run chatbox.
   Make sure that ollama is running before you run chatbox.（See step of "Start ollama"）

   ./Chatbox-1.9.8-x86_64.AppImage --no-sandbox

3. Click setting
   Edit API Host to http://localhost:11434/
   Select your model, edit your max message count, change temperature and save
   
4. At last click New Chat to start a new communication
   

References

https://github.com/ollama/ollama
https://ollama.com/library/deepseek-r1