Status: [UNTESTED] Endpoints found: http://127.0.0.1:18790/api/dry-run/enable http://127.0.0.1:18790/api/disable` Source: https://github.com/Ath404/ClawRoute # 🦀 ClawRoute **Intelligent cost-optimizing model router for OpenClaw** [Website](https://clawroute.atharvc.tech) · [npm](https://www.npmjs.com/package/clawroute) Save 60-90% on your LLM costs by automatically routing requests to cheaper models when appropriate. > **v1.3** — Intelligent, local, and private. Updated for the latest models (Oct 2026), with native Anthropic support. > > Works with **any OpenAI-compatible client** — OpenClaw, Cursor, custom apps, or direct API calls. No special detection needed. ## What's New in v1.3 - **Latest models**: Claude Haiku 5.5 / Sonnet 5.5 / Opus 5.5 / Fable 5.1, GPT-6 Luna / 6.1 Sol / Astra, Gemini 3.5 Flash Lite / 3.8 Flash / 4 Argon. Default tiers updated; older models stay in the registry for cost lookups. - **Native Anthropic adapter**: direct `anthropic/*` routing, including streaming and tool use (no OpenRouter needed). - **Fix**: the bundled default config now ships in the npm package. If you customised tiers in `config/clawroute.json`, your settings still take precedence over the new defaults. ## How It Works ClawRoute is a local HTTP proxy that sits between OpenClaw and your LLM providers. It: 1. **Intercepts** every LLM request from OpenClaw 2. **Classifies** the task complexity using local heuristics (no API calls, <5ms) 3. **Routes** to the cheapest model that can handle it 4. **Escalates** automatically if the cheap model fails (before streaming) ``` OpenClaw → ClawRoute (127.0.0.1:18790) → OpenAI/Anthropic/Google/DeepSeek ↓ [Classify → Route → Execute] ↓ SQLite logs (savings tracked) ``` ## Quick Start ### From npm ```bash npm install -g clawroute export OPENAI_API_KEY=sk-xxx # or put your keys in a .env where you run it clawroute start # proxy on http://127.0.0.1:18790 ``` ### From source ```bash # Clone and install git clone https://github.com/Ath404/ClawRoute && cd ClawRoute npm install && npm run build # Con