Which AI models are supported?

ClawRouter supports over 30 models including GPT-4o, Claude 3.5 Sonnet, Gemini 2.0 Pro, DeepSeek Chat, and Grok-3.

How do I enable smart routing?

Once installed, you can enable autonomous routing by setting your model to 'blockrun/auto' within your Claude Code configuration.

Does ClawRouter increase response latency?

Minimally. Approximately 80% of requests are handled by rule-based classification in less than 1ms. Ambiguous queries use a micro-LLM classifier that adds negligible cost and time.

How much can I save using ClawRouter with Claude Code?

Users typically save around 78% on inference costs by routing simpler tasks to low-cost models like Gemini Flash and DeepSeek while reserving premium models for complex reasoning.

Does this require crypto payments?

No, this version of ClawRouter is a specialized fork that works with your direct API keys or OpenRouter, eliminating the need for crypto-based wallet systems.

ClawRouter

Name: ClawRouter
Author: cgaeking

bycgaeking

•

데이터 과학 및 ML

Reduces LLM inference costs by automatically routing requests to the most cost-effective and capable models across major providers.

ClawRouter is a specialized optimization skill for Claude Code designed to slash AI development budgets by up to 78%. By analyzing incoming requests and classifying them into four complexity tiers—Simple, Medium, Complex, and Reasoning—it dynamically routes each task to the cheapest model capable of handling it. Whether you are performing basic translations with Gemini Flash or complex logic with Claude Opus, ClawRouter ensures you never overpay for inference while maintaining peak performance across 30+ models from OpenAI, Anthropic, Google, and DeepSeek.

주요 기능

01Support for 30+ models from 5 major providers via direct API keys or OpenRouter

02High-speed rule-based classification with sub-1ms latency for 80% of requests

03Real-time cost savings reporting and model confidence tracking

04Seamless switching between pinned models and autonomous smart-routing

0518 GitHub stars

06Smart multi-tier routing based on task complexity and reasoning requirements

사용 사례

01Offloading high-volume, low-complexity tasks to hyper-efficient models like Gemini and DeepSeek

02Benchmarking performance across multiple LLM providers through a unified interface

03Scaling AI-powered applications while strictly managing operational expenses

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add cgaeking/clawrouter clawrouter

For use in Claude.ai and ChatGPT

Download Skill