Gemini CLI - OpenRouter Edition

🚀 Enhanced Gemini CLI with OpenRouter Support - Access 200+ AI Models Through One Interface

⚠️ Important Notice: This is a modified version of Google Gemini CLI. Original project copyright belongs to Google LLC under Apache 2.0 License.

🌟 Project Status & Future Development

Current Status: This project currently supports a curated selection of high-performance models through OpenRouter integration. We're actively working to expand model compatibility and enhance features.

Community Driven: Your support helps us grow! ⭐ Star this repository and consider contributing to help us:

• Add support for more AI models
• Improve performance and reliability
• Develop new features and integrations
• Maintain comprehensive documentation

Get Involved: Have suggestions or want to contribute? Contact us at mythicscribe2014@gmail.com - we'd love to hear from you!

🌐 Language Switch

| 🇺🇸 English | 🇻🇳 Tiếng Việt | 🇯🇵 日本語 | 🇩🇪 Deutsch | 🇸🇦 العربية | 🇨🇳 中文 |

🌟 Core Features

• 🎯 OpenRouter Integration: Access 200+ AI models through OpenRouter's unified API
• 🔄 Model Flexibility: Switch between GPT-4, Claude, Llama, Mistral, and many more
• 🌍 Global Model Coverage: Supports all major AI models from OpenAI, Anthropic, Meta, Google, and more
• 🔧 Simple Configuration: Single API key for all models
• 📊 Complete Functionality: Streaming/non-streaming, token counting, function calling all supported
• 🚀 Enhanced Performance: Optimized for OpenRouter's high-performance infrastructure

📦 Installation

🚀 Quick Install (Recommended)

npm install -g @chameleon-nexus-tech/gemini-cli-openrouter

📋 Alternative Installation Methods

# Using npm (Node.js package manager)
npm install -g @chameleon-nexus-tech/gemini-cli-openrouter

# Using yarn
yarn global add @chameleon-nexus-tech/gemini-cli-openrouter

# Using pnpm
pnpm add -g @chameleon-nexus-tech/gemini-cli-openrouter

🎛️ OpenRouter Integration

🌐 OpenRouter - Your Gateway to 200+ AI Models

OpenRouter provides access to the world's most advanced AI models through a single, unified API. This enhanced Gemini CLI leverages OpenRouter to give you instant access to:

🔥 Popular Models Available:

• anthropic/claude-3.5-sonnet (default) - Anthropic's most capable model
• openai/gpt-4 - OpenAI's flagship model
• openai/gpt-4-turbo - Faster GPT-4 variant
• meta-llama/llama-3.1-8b-instruct - Meta's latest Llama model
• meta-llama/llama-3.1-70b-instruct - Meta's most powerful Llama
• mistralai/mistral-7b-instruct - Mistral's efficient model
• mistralai/mixtral-8x7b-instruct - Mixtral's mixture of experts
• google/gemini-pro - Google's Gemini model
• google/gemini-pro-vision - Google's multimodal model
• cohere/command-r-plus - Cohere's advanced model
• deepseek/deepseek-coder - Specialized coding model
• And 190+ more models from leading AI companies!

🎯 Why OpenRouter?

• Single API Key: Access all models with one credential
• Best Pricing: Competitive rates across all providers
• High Availability: 99.9% uptime with global infrastructure
• Model Comparison: Easy switching between models for testing
• Enterprise Features: Rate limiting, usage analytics, and more

⚙️ Configuration

🔧 Required Environment Variables

You only need to set these 4 environment variables to get started:

# 1. Set AI engine to OpenRouter
export AI_ENGINE="openrouter"

# 2. Set your OpenRouter API key (get from https://openrouter.ai/keys)
export AI_API_KEY="your-openrouter-api-key"

# 3. Choose your preferred model
export AI_MODEL="openai/gpt-4o"

# 4. Set any value to skip Gemini API validation
export GEMINI_API_KEY="any-value-here"

🪟 Windows PowerShell Configuration

# Set the 4 required environment variables
$env:AI_ENGINE="openrouter"
$env:AI_API_KEY="your-openrouter-api-key"
$env:AI_MODEL="openai/gpt-4o"
$env:GEMINI_API_KEY="any-value-here"

🎯 Quick Setup

That's it! These 4 variables are all you need:

• AI_ENGINE="openrouter" - Tells the system to use OpenRouter
• AI_API_KEY="your-key" - Your OpenRouter API key
• AI_MODEL="model-name" - The AI model you want to use
• GEMINI_API_KEY="anything" - Set to any value to skip validation

🚀 Usage

💬 Direct Question Mode

gemini "Hello, please introduce yourself"

🔄 Interactive Mode

gemini

Then select "2. Use Gemini API Key" and press Enter to start chatting.

🧪 Model Testing Examples

🌐 Testing with GPT-4o

# Set the 4 required environment variables
export AI_ENGINE="openrouter"
export AI_API_KEY="your-openrouter-api-key"
export AI_MODEL="openai/gpt-4o"
export GEMINI_API_KEY="any-value-here"

# Test command
gemini "Hello, please introduce yourself and tell me which AI model you are."

🏗️ Technical Architecture

🎯 OpenRouter Integration Architecture

This project features a specialized OpenRouter integration that provides seamless access to 200+ AI models:

OpenRouter Integration
├── OpenrouterContentGenerator    (Primary OpenRouter adapter)
├── ContentGeneratorFactory       (Engine selection and management)
├── Tool Execution System        (Function calling support)
└── Streaming Response Handler    (Real-time response processing)

🔧 Core Features

✅ OpenRouter Integration: Direct access to 200+ AI models through OpenRouter's unified API
✅ Model Flexibility: Easy switching between different AI models with a single environment variable
✅ Automatic Format Conversion: Seamless conversion between Gemini format and OpenRouter API format
✅ Streaming Response Support: Real-time streaming responses for all supported models
✅ Function Calling: Full support for tool/function calling across all models
✅ Intelligent Error Handling: Comprehensive error handling with detailed messages
✅ Configuration Priority: OpenRouter-specific config > Unified config > Default config

📁 Core File Structure

packages/core/src/core/
├── contentGenerator.ts              # ContentGenerator interface definition
├── contentGeneratorFactory.ts       # Engine factory with OpenRouter priority
├── openrouterContentGenerator.ts    # OpenRouter adapter (primary)
├── coreToolScheduler.ts            # Tool execution and function calling
└── [other adapters]                 # Additional engine support

🔄 OpenRouter API Integration

Request Conversion Flow:

Gemini Request → OpenRouter Adapter → OpenRouter API → 200+ AI Models

Response Conversion Flow:

AI Model Response → OpenRouter API → OpenRouter Adapter → Gemini Response Format

Supported OpenRouter Features:

• Model Selection: Easy switching between 200+ models
• Function Calling: Full tool/function calling support
• Streaming: Real-time response streaming
• Token Counting: Automatic token usage tracking
• Error Handling: Comprehensive error management
• Rate Limiting: Built-in rate limiting and retry logic

📋 License Compliance

This project is based on Google Gemini CLI and follows Apache 2.0 License requirements:

Original Project Information

• Original Project: Google Gemini CLI
• Original Copyright: Copyright 2025 Google LLC
• Original License: Apache License 2.0
• Original Repository: https://github.com/google-gemini/gemini-cli

Modification Statement

In accordance with Apache 2.0 License Section 4:

• ✅ Retained original Apache 2.0 license
• ✅ Retained original copyright notices
• ✅ Clearly marked modifications
• ✅ Included complete license text

🔄 Migration from Original Gemini CLI

If you're migrating from the original Google Gemini CLI:

1. Uninstall original version: npm uninstall -g @google/gemini-cli
2. Install this version: npm install -g @chameleon-nexus-tech/gemini-cli
3. Get OpenRouter API key: Sign up at OpenRouter
4. Set the 4 required environment variables:

   export AI_ENGINE="openrouter"
   export AI_API_KEY="your-openrouter-api-key"
   export AI_MODEL="openai/gpt-4o"
   export GEMINI_API_KEY="any-value-here"

5. Use as before: All commands remain the same, now with access to 200+ models!

🧪 Testing OpenRouter Integration

🚀 Quick Test Script

Create a simple test script to verify your OpenRouter setup:

#!/bin/bash
# test-openrouter.sh

# Set the 4 required environment variables
export AI_ENGINE="openrouter"
export AI_API_KEY="your-openrouter-api-key"
export AI_MODEL="openai/gpt-4o"
export GEMINI_API_KEY="any-value-here"

echo "Testing OpenRouter with GPT-4o..."
gemini "Hello, please introduce yourself and tell me which AI model you are."

🔧 Model Comparison Test

Test different models to compare their responses:

#!/bin/bash
# compare-models.sh

# Set up the base configuration (set once)
export AI_ENGINE="openrouter"
export AI_API_KEY="your-openrouter-api-key"
export GEMINI_API_KEY="any-value-here"

# Test GPT-4o
echo "=== Testing GPT-4o ==="
export AI_MODEL="openai/gpt-4o"
gemini "Write a haiku about artificial intelligence."

🔗 Related Links

• Original Project (Google Gemini CLI)
• OpenRouter - Access 200+ AI models
• OpenRouter API Documentation
• OpenRouter Model List
• Apache 2.0 License

🐛 Issue Reporting

If you encounter any issues, please report them on GitHub Issues.

🤝 Contributing

Contributions are welcome! Please ensure:

1. Follow the original project's code style
2. Include appropriate tests
3. Update relevant documentation
4. Respect the Apache 2.0 license terms

🌟 Project Highlights

• 🚀 OpenRouter Integration: Direct access to 200+ AI models through one API
• 🎯 Model Flexibility: Easy switching between GPT-4, Claude, Llama, and more
• 🔄 Simple Configuration: Single API key for all models
• 🌍 Global Coverage: Access to models from OpenAI, Anthropic, Meta, Google, and more
• 📊 Enterprise Grade: Complete error handling and configuration management
• 🔧 Function Calling: Full support for tool/function calling across all models

---

Disclaimer: This project is not affiliated with Google or OpenRouter. Please comply with the terms of service of the respective platforms when using this software.