BlockRunAI/ClawRouter

The agent-native LLM router for autonomous agents. Every frontier model behind one wallet, <1ms local routing, USDC payments on Base & Solana via x402.

What it solves

Autonomous AI agents cannot sign up for accounts, enter credit cards, or manage API keys like humans do. ClawRouter solves this by letting agents pay per request with USDC using a wallet signature as authentication, while also offering a traditional credit-card-funded API key for humans. It also reduces LLM API costs significantly by automatically choosing the cheapest model that can handle each request.

How it works

ClawRouter runs as a local proxy (port 8402) and uses an open-source routing engine that scores every request across 15 dimensions (such as complexity, context length, and modality) to pick the best model from a pool of over 90 across 9 providers. Routing decisions happen entirely on your machine with under 1ms overhead and zero external calls. Payment is flexible: either a locally generated wallet signs per-request USDC payments on Solana or Base via the x402 protocol, or a human can sign up, add credit with a card, and mint an API key. A free tier of 7 open-weight models requires no payment at all.

Who it’s for

Primarily autonomous AI agents (like OpenClaw, Codex, Hermes) that need to pay for API usage without human intervention, and developers who want smart routing, cost savings, and the option to pay via card. It also suits anyone who wants to avoid managing multiple API keys and model selections.

Highlights

  • 100% local routing with less than 1ms latency and no external API calls
  • Routes across 15 dimensions to the cheapest capable model, cutting costs by up to 84% vs. pinning one premium model
  • Supports USDC micropayments on Solana and Base via x402, plus a credit-card-funded API key option
  • 7 free models with no signup, key, or wallet required
  • Integrates as a plugin for OpenClaw with one-shot installer, and works with any OpenAI-compatible client as a local proxy
  • Open source (MIT) and agent-native: wallet signature is the only authentication

Related

  • Project
    Continuum-AI-Corp/OrcaRouter-Lite

    A self-hosted, OpenAI-compatible LLM router that automates model selection based on cost and capability, providing a unified API for multiple AI providers.

  • Dispatch
    Weave Router: Intelligent Model Routing for Agentic Systems

    Weave Router is a drop-in proxy that reduces LLM costs by 40-70% by routing prompts to the most efficient model in under 50ms using a cluster scorer derived from Avengers-Pro.

  • Project
    decolua/9router

    9Router is a Node.js/Next.js service that acts as a local OpenAI‑compatible gateway for AI coding tools. It compresses payloads (RTK Token Saver), tracks quota, and automatically falls back across paid, cheap, and free LLM providers, supporting 40+ APIs and self‑hosted servers. The goal is to keep coding uninterrupted while cutting token costs by 20‑40 %.

  • Project
    TokenFlux/TokenRouter

    An open-source unified gateway for large language models that consolidates multiple AI providers into a single API entry point with built-in billing and access control.

  • Project
    NadirRouter/NadirClaw

    NadirClaw is a local, source-available LLM router that reduces API costs by intelligently routing prompts to the cheapest capable model and escalating only when necessary.