Roo Code Integration Guide
This guide will help you integrate GPT-Load proxy service into Roo Code intelligent code assistant, supporting OpenAI, Gemini, and Anthropic channel types.
Prerequisites
Ensure you have successfully deployed and started the GPT-Load service, running by default at http://localhost:3001
1OpenAI Channel Configuration
Configuration Steps
Select Provider
Select "OpenAI Compatible" provider in Roo Code
Set Proxy Address
http://localhost:3001/proxy/openai/v1
Where "openai" is the group name you configured in GPT-Load
Configure API Key
Enter the proxy key configured in GPT-Load

Click image to enlarge
2Gemini Channel Configuration
Method 1: Google Gemini Provider
Using native Gemini interface format (recommended)
Configuration Steps
Select Provider
Select "Google Gemini" provider in Roo Code
Enable Custom Base URL
Check "Use custom base URL" option
Set Proxy Address
http://localhost:3001/proxy/gemini
Where "gemini" is the group name you configured in GPT-Load
Configure API Key
Enter your Gemini API Key

Click image to enlarge
Method 2: OpenAI Compatible Format
Using OpenAI compatible interface format
Configuration Steps
Select Provider
Select "OpenAI Compatible" provider in Roo Code
Set Proxy Address
http://localhost:3001/proxy/gemini/v1beta/openai
Where "gemini" is the group name you configured in GPT-Load
Configure API Key
Enter your Gemini API Key

Click image to enlarge
Configuration Recommendation
We recommend using Method 1: Google Gemini Provider, as it uses the native Gemini interface format, providing better compatibility and complete feature support.
3Anthropic (Claude) Channel Configuration
Configuration Steps
Select Provider
Select "Anthropic" provider in Roo Code
Set Proxy Address
http://localhost:3001/proxy/anthropic
Where "anthropic" is the group name you configured in GPT-Load
Configure API Key
Enter your Anthropic API Key

Click image to enlarge
Verify Configuration
Test Connection
After configuration is complete, send a test message in Roo Code to confirm normal response. If you encounter issues, please check if the GPT-Load service is running normally and if the group name is correct.