The entire process is free and practical, and you can experience DeepSeek V4 Flash access to Codex without paying.
This tutorial takes you through the entire process from Codex installation, free credit acquisition, to DeepSeek access configuration.
Suitable for the crowd:
- Developers who want to use more AI models in Codex
- Users who want to experience DeepSeek V4 Flash
- Friends who want to use AI programming assistants at low cost
The entire process requires no complex operations, and only needs to follow the steps to complete the configuration.
1. Install Codex
First install Codex through a Windows terminal.
1. Open the administrator terminal
Right-click on the Windows Start menu and select:
Terminal (Administrator)
Enter the following command in the pop-up window:
winget.exe install --id 9NT1R1C2HH7J --source msstore --accept-package-agreements --accept-source-agreements --silent
Just wait for the installation to complete.
After installation is complete:
- press the Windows key
- Search and open ChatGPT
- Complete basic setup
At this point, the Codex installation is complete.
2. Get free credit for DeepSeek V4 Flash
The free model credit provided by OpenCode Zen is used here.
The principle is simple:
You can call DeepSeek V4 Flash by registering an OpenCode account, obtaining the OpenCode Zen API Key, and then connecting the interface to Codex.
1. Sign up for OpenCode account
Open the registration page:
Complete registration and log in.
After entering the Zen package page, you can create the corresponding API Key.
Note: API Key is a personal key and do not share it publicly.
After completing the above steps, you will receive a free credit for DeepSeek V4 Flash.
2. API configuration information
If you want to use it on other platforms that support the OpenAI Compatible protocol, you can refer to the following configuration:
API Key：
Fill in the OpenCode Zen API Key you just created
Base URL：
https://opencode.ai/zen/v1
Model name:
deepseek-v4-flash-free
Interface protocol:
OpenAI Compatible / Chat Completions
3. Connect DeepSeek V4 Flash to Codex
Cockpit Tools is used here as a configuration management tool.
It helps Codex quickly switch between different AI services while retaining information about the local project environment, context, and more.
1. Download Cockpit Tools
GitHub download address:
You can install directly after downloading.
2. Create a Codex configuration
Open Cockpit Tools:
- left selection Codex
- click on the right + create a configuration
- Select in the pop-up window API Key
Then fill in the following:
supplier type
Choose:
自定义
API Key
Copy the previously saved OpenCode Zen API Key.
base address
Fill in:
https://opencode.ai/zen/v1
API supplier name
You can fill in freely, such as:
DeepSeek Zen
model list
Fill in:
deepseek-v4-flash-free
Then:
- Check "Sync suppliers to Codex"
- Click "Add"
3. Adjustment model protocol
After the addition is completed:
- Open "Model Supplier" above
- Find the supplier you just created
- click Edit
Confirm the following settings:
Collaboration ability:
Chat Completions
List of models:
deepseek-v4-flash-free
Click Save after confirming that there is no error.
Finally:
Click the supplier launch button and wait for the launch to complete.
Now Codex has successfully connected to DeepSeek V4 Flash.
4. Usage effects and extended gameplay
After configuration, you can directly call DeepSeek V4 Flash in Codex.
At the same time, this method is not limited to DeepSeek:
By modifying the API address and model name, you can also access other models that are compatible with the OpenAI API, such as:
- Grok
- GLM
- Kimi
- Other third-party AI models
Cockpit Tools supports saving multiple vendor configurations and can be switched at any time as needed.
More importantly:
- Does not affect local Codex projects
- Does not affect existing context
- Can manage multiple AI models simultaneously
summary
Through the above steps, you can:
Install Codex for free
Get DeepSeek V4 Flash usage credits
Connect DeepSeek to Codex
Use Cockpit Tools to manage multiple AI model configurations
The entire process only needs to be configured once, and then you can freely switch between different models according to your needs, giving Codex more powerful AI capabilities.
I hope this tutorial can help more developers experience AI programming tools at a low cost.