Tencent's CodeStudio Deploys the Deepseek-R1 Large Model with Ollama#
Tencent's CodeStudio can actually deploy the Deepseek-R1 large model using Ollama. The key is the 10,000 minutes quota per month, which is practically free! Friends with ordinary computers can register for Tencent Cloud to personally experience the joy of deploying a large model.
Preparation Work#
Ollama is a free open-source tool that allows users to run large language models (LLMs) locally on their computers. It is compatible with macOS, Linux, and Windows. In simple terms: it is a handy assistant that allows you to install, deploy, and run large models with just one command.
Official Address: Ollama Official Website
If you want to deploy a large model on your own computer, this tool is a must-have!
Steps for Cloud Deployment#
Without further ado, let's get started!
-
Visit Tencent Cloud: Tencent Cloud IDE
-
Select Space Template: Choose the "ollama" template in the "Space Template" section. It will prompt you with 10,000 minutes of free time each month, which is 166.7 hours.
. Ollama List Appears: At this point, an Ollama list will appear.
4. Open Terminal: A VSCode dialog box will appear; follow the steps in the screenshot to open a terminal area.
5. Input Command: Directly enter the code:
ollama run deepseek-r1
Press the "Enter" key to execute. It will pull the image from the remote repository, and by default, it will download the cloud large model with 7B parameters.
In no time, it will be installed, and a >>> prompt will appear, allowing us to converse with Deepseek-R1.
Find the port where the Ollama service is running
It is found to be 6399. Since CodeStudio comes with a VSCode code development window, we can also use Tencent's built-in "AI Code Assistant" to generate a program that calls "deepseek-r1".
Specific Command
It can be written like this:
curl http://0.0.0.0:6399/api/chat -d '{
"model": "deepseek-r1",
"messages": [
{ "role": "user", "content": "Hello" }
]
}'
Then you can easily call the Deepseek API, making development and debugging convenient.
Note
Although 10,000 minutes are given each month, running 24 hours a day will only last for a little over a week. If you run out, you can return to the homepage and simply click to shut down.