Precautions
1、Please make sure your hardware meets the system requirements to get the best performance and user experience.
2、Please make sure that you have fully read the license of DeepSeek and use DeepSeek model in strict accordance with the license agreement.

DeepSeek mirroring service is only for demonstration and test use, please do not use this service for illegal purposes, Chengdu Cloud Nine Days Network Technology Co. You must bear the risks and consequences of using the mirror. Before using the mirror, please ensure that you have a detailed understanding of the contents and have the appropriate technical skills and knowledge. It is recommended that you use it in a test environment and in compliance with the relevant license agreements and laws and regulations.

Please note that this image is intended for personal study, testing and research only, and is not recommended for use in a production environment. For any problems caused by the use of this mirror in the production environment, Chengdu Cloud Jiu Tian Network Technology Co.
Use of this mirror means that you agree to bear all the risks and responsibilities.

1. Introduction to DeepSeek

DeepSeek-R1 is an open source AI large-scale language model released on January 20, 2025 by DeepSeek. Performance on math, code and reasoning tasks is comparable to OpenAI-o1. DeepSeek-R1(7b) is a mid-range version of DeepSeek R1. Its design goal is to transfer the powerful inference capabilities of DeepSeek R1 to smaller models for improved computational efficiency and deployment flexibility.

System Requirements:
It is recommended that your ecs host meets the following requirements:

1.5b model: need about 1GB video memory, RAM 8GB or more.
7b/8b model: need 4GB video memory, memory 16GB or more.
14b/32b model: 8GB/18GB of video memory, 32GB of RAM or more.
70b/671b models: 40GB+ video memory, 64GB+ RAM.

2. Product introduction

①.based on AlibabaCloudLinux 3.2LTS install Ollama platform, download the big models (LLMs), and through the Docker installed the web visualization service OpenWebUI, port 8080 ;.

②.Ollama can deploy and run large models (LLMs) faster. OpenWebUI facilitates users to access and interact with large models through the browser to enhance the experience.

③.WebUI uses port 8080, before using it, please release port 8080 in the instance association security group as shown. For more information, please refer to the tutorial https://www.alibabacloud.com/help/en/elastic-compute-service/latest/add-security-group-rules

3.How to use

After purchase, please wait about 5-10 minutes for the system to finish deploying, and then enter http://public_ip:8080/ to access.

Please initialize your account for the first time.

4. Local use

①., if you locally installed ChatBox, CherryStudio and other clients, also supports remote call, you need to first release the port 11434 in the security group, according to the above security group release instructions.
②.Local client configuration address is to use http://public_ip:11434 

5. Service Description

    Email: admin@yunjiutian.com on working days
    Service time: 5 * 8 hours on working days

Provide purchase consultation, installation guidance, image reinstallation, and no free technical support for non image problems .