Claude Code Router + Gemini 2.5 Pro FREE API: RIP Gemini CLI! This is what Gemini-CLI should be!

Key Concepts

Claude Code, Claude Code Router, Anthropic models, Gemini CLI, Codeex CLI, OAMA, Background model, Think model, Long context model, Plugins, Ninja Chat, Model configuration, API key, CI workflows.

Claude Code Router: Connecting Claude Code to Any Model

The video discusses Claude Code Router, a tool that allows users to connect Claude Code to various models beyond Anthropic's, addressing the limitations of Claude Code being restricted to its own (often expensive) models and not being open source. This opens up possibilities to use models like Gemini, Mistral, DeepSeek, or even local models via OAMA.

Model Configuration: Setting Up the Models

Configuring Claude Code Router involves setting up four different models for different tasks:

1. Background Model: Handles tasks like haiku generation for loading messages, conversation summarization, and command processing. The speaker recommends using Gemini 2.0 Flash or a small local model for this.
2. Think Model: Used when Claude Code is asked to "think," acting as the reasoning model. The speaker uses Gemini 2.5 Pro for this, also used as the general model.
3. Long Context Model: The speaker uses the same model as the Think model and general model.
4. General Model: The speaker uses the same model as the Think model and long context model.

The speaker highlights that Gemini 2.5 Pro offers 1,500 free requests per day via its API, making it a cost-effective option for the Think and General models.

Plugins: Enhancing Model Performance

Plugins are used to modify requests, adding system prompts to improve model performance. The speaker mentions a plugin that works well with DeepSeek models.

Installation and Usage: A Step-by-Step Guide

1. Installation: Install Claude Code Router using the command provided in the video.
2. Starting the Router: Run CC start to initiate the router. This will prompt you to configure the main model.
3. Configuration: Enter the base URL, API key, and model name for your chosen model (e.g., Gemini). The configuration can be customized to use different models for different tasks (background, think, long context, general).
4. Using Claude Code: Run the CCR code command to start Claude Code with the router. This will use the configured model for all Claude Code operations.

The speaker notes that delays in responses might occur due to the rate limits of free tiers like Gemini's.

Example: Creating a Minesweeper Game

The speaker demonstrates the functionality by asking Claude Code to create a Minesweeper game using HTML, CSS, and JS. The process mirrors the standard Claude Code experience, with the underlying model (Gemini in this case) handling the code generation. The speaker notes that the tool works well with Gemini, and the tools and MCPS work fine.

Addressing Potential Issues

The speaker acknowledges that Claude Code Router can sometimes produce errors, particularly with diff editing, but these are rare and can be mitigated using plugins.

Integration with CI Workflows

The speaker suggests that Claude Code Router can be integrated with CI workflows using tools like GitHub Actions.

Ninja Chat: An All-in-One AI Platform

The video includes a brief advertisement for Ninja Chat, an AI platform offering access to models like GPT-4o, Claude 3.7 Sonnet, and Gemini 2.0 Flash for $11 per month. Key features include:

• AI Playground: Allows side-by-side comparison of responses from different models.
• Mind Map Generator: Helps organize complex ideas.
• Pricing: The basic plan includes 1,000 messages, 30 images, and 5 videos monthly.
• Discount Codes: The speaker provides discount codes "king25" for 25% off any plan and "king40yearly" for 40% off annual subscriptions.

Conclusion

Claude Code Router provides a valuable solution for users seeking to leverage Claude Code's interface and tools with a wider range of models, including open-source and cost-effective options. While configuration requires setting up multiple models for different tasks, the flexibility and potential cost savings make it a compelling alternative to being locked into Anthropic's ecosystem. The speaker emphasizes its seamless integration and functionality, making it a worthwhile tool for AI-assisted coding and other tasks.