Verified project record

BlockRunAI/ClawRouter

An open-source LLM router enabling autonomous AI agent operation through local execution, wallet signature authentication, and USDC micropayments. It routes requests to models in under 1ms and supports text, image, video, and audio modalities without mandatory signup.

Project overview

Project type: Model Runtime · Infrastructure
Use cases: Automation
Deployment: Refer to project documentation
License: MIT

Best for

Developers building AI agents that require autonomous payment processing and local model routing without external API dependencies.
Engineers looking to evaluate a local LLM routing implementation operating under 1ms.

Key capabilities

Analyzes each request across 15 dimensions and routes to the cheapest capable model locally in under 1ms.
Provides access to 8 free NVIDIA models out of the box without requiring crypto or signup.
Proxies image generation requests via chat commands or HTTP API.
Proxies video generation requests, handling payments and downloading the returned MP4 to local disk.
Ships a skill allowing agents to query BlockRun's unified crypto data API with 84 endpoints across 13 domains.

Limitations and risks

Using models outside the free tier incurs real USDC costs and requires interaction with the Base (EVM) or Solana networks.

Getting started

Run via a simple npx command without requiring signup or coding, then point your OpenAI-compatible client at the local instance.

Evidence and sources

GitHub project description: The agent-native LLM router for OpenClaw. 41+ models, <1ms routing, USDC payments on Base & Solana via x402.
README: Agents can't do any of that.
README: ClawRouter is the only LLM router that lets agents operate independently.
README: It analyzes each request across 15 dimensions and routes to the cheapest capable model in under 1ms, entirely locally.
README: 100% local routing. <1ms latency. Zero external API calls.

Reviewed at: 2026-07-15T15:06:41Z