Gemini CLI Integration with RezeAI Tutorial

Google's official Gemini CLI is an extremely powerful endpoint AI tool (with a built-in, highly praised interactive interface). Through simple environment variable configuration, you can seamlessly switch its default cloud computing power to the high-performance model services provided by RezeAI.

I. Preparation Before Integration (3 Core Elements)

Before starting, please be sure to clarify the following 3 parameters. Please pay special attention to the format requirements of the Base URL:

| Configuration Items | Recommended Content | Notes |

| :--- | :--- | :--- |

| API Key | sk-xxxxxxxxxxxxxxxx | The plain text key obtained from the RezeAI console, do not include the Bearer prefix. |

| Base URL | https://api.rezeai.com | ⚠Cannot include /v1**. Because Gemini CLI automatically appends the official path to the end. | | Model Name | Enter the name of a Gemini model supported by RezeAI | It is recommended to use a natively supported model name such as gemini-2.5-flash-thinking. |

II. Installing Gemini CLI

Gemini CLI depends on a Node.js environment. Please ensure that Node.js is installed on your computer.

If your environment does not have Node.js, click to download

Open a terminal (Windows recommends PowerShell, Mac recommends Terminal), and run the following official command for global installation:

III. Configuring Environment Variables (Core Step)

Gemini CLI does not recognize generic OpenAI environment variable names; you must use its proprietary variable names (GEMINI_API_KEY, GOOGLE_GEMINI_BASE_URL, etc.).

Please copy and modify the following commands according to your operating system (replace with your own API Key and model name), and run them in the terminal:

Windows (PowerShell):


$env:GEMINI_API_KEY="sk-your Reze key"

$env:GOOGLE_GEMINI_BASE_URL="https://api.rezeai.com"

$env:GEMINI_MODEL="gemini-2.5-flash-thinking"

Windows (CMD):


set GEMINI_API_KEY=sk-your Reze key

set GOOGLE_GEMINI_BASE_URL=https://api.rezeai.com

set GEMINI_MODEL=gemini-2.5-flash-thinking

macOS / Linux:

💡 Advanced Tips: The above environment variables are only valid for the current terminal window by default. To make them permanent, add them to your system's "Advanced System Settings -> Environment Variables", or write them to your Mac/Linux's ~/.bashrc / ~/.zshrc configuration file.

IV. One-Click Startup and First Boot

After injecting the environment variables, enter the following command in the same terminal window to start:

Possible prompts on first boot:

Install VS Code plugin: If prompted Do you want to connect VS Code to Gemini CLI?, it is strongly recommended to press 1 (Yes). This allows AI to directly read the code in your editor.

Passwordless Login: Because you have correctly configured your environment variables, the program will automatically skip the login menu and directly enter the dialog interface with the ">" prompt.

After the ">", type Hello, please introduce yourself. If you receive a reply quickly, congratulations, the integration is successful!

V. Common Problem Troubleshooting (FAQ)

Q1: Error Invalid URL (POST /v1/v1beta/models/...:streamGenerateContent)

Cause: This is the most common error! It means that your GOOGLE_GEMINI_BASE_URL is entered as https://api.rezeai.com/v1.

Solution: Please delete /v1 at the end of the URL. It must remain the clean root domain https://api.rezeai.com, then restart the terminal to test.

Q2: After starting, a bunch of red text pops up: [API Error: Content generator not initialized], and it forcibly prompts you to select an authentication method?

Reason: This indicates that Gemini CLI has not read your environment variables.

Solution:

Please check if you mistakenly wrote OPENAI_API_KEY as the variable name (it must be GEMINI_API_KEY).

Check if you accidentally closed the terminal after configuring the environment variables. Please run gemini in the same window where you configured the environment variables.

Q3: The dialog prompts that the model cannot be found or is invalid?

Solution: Please check if the model name entered in the environment variable $env:GEMINI_MODEL is correct. It is recommended to use the full name of a real Gemini series model within the RezeAI platform for testing.

Gemini CLI

Gemini CLI Integration with RezeAI Tutorial#

I. Preparation Before Integration (3 Core Elements)#

II. Installing Gemini CLI#

III. Configuring Environment Variables (Core Step)#