SimpleClawRouter

Save 10x on LLM costs. Automatically.

Route every request to the cheapest model that can handle it with ultra-fast local regex routing using your OpenRouter key.

This is a fork of BlockRun's ClawRouter, a weighted regex classification based local router. This fork replaces the x402 crypto-wallet payment gateway with OpenRouter, so you can use a standard API key instead of managing USDC wallets. If crypto x ai excites you, go check out the original repo! For a more intelligent routing endpoint, you can use openrouter/auto (powered by Not Diamond), a general-purpose model router built directly into OpenRouter."

---

"What is 2+2?"            → Gemini 2.5 Flash Lite $0.10/M    saved 99%
"Summarize this article"  → Gemini 3 Flash       $0.50/M    saved 98%
"Build a React component" → Claude Sonnet 4.5    $3.00/M    best balance
"Prove this theorem"      → Gemini 3 Pro         $2.00/M    reasoning
"Run 50 parallel searches"→ Kimi K2.5            $0.45/M    agentic swarm

Why SimpleClawRouter?

• 100% local routing — 14-dimension weighted scoring runs on your machine in <1ms
• 16 models — OpenAI, Anthropic, Google, DeepSeek, xAI, Moonshot, MiniMax, Z.AI, Arcee via OpenRouter
• Standard API key — just set OPENROUTER_API_KEY and go
• Open source — MIT licensed, fully inspectable routing logic

---

Quick Start

# 1. Install
openclaw plugins install clawrouter-openrouter

# 2. Set your OpenRouter API key
export OPENROUTER_API_KEY=sk-or-...
# Get your key at https://openrouter.ai/keys

# 3. Restart OpenClaw to load the plugin
openclaw restart

To enable smart routing, add to ~/.openclaw/openclaw.json:

{
  "agents": {
    "defaults": {
      "model": {
        "primary": "clawrouter/auto"
      }
    }
  }
}

Or use /model clawrouter/auto in any conversation to switch on the fly.

Want a specific model? Use openai/gpt-5.2 or anthropic/claude-opus-4.6 directly — requests still go through the proxy with dedup and streaming optimizations.

---

How Routing Works

100% local, <1ms

Request → Weighted Scorer (14 dimensions)
              │
              ├── High confidence → Pick model from tier → Done
              │
              └── Low confidence → Default to MEDIUM tier → Done

Routing is done through regex + weighted multi-class classification:

14-Dimension Weighted Scoring

Dimension	Weight	What It Detects
Reasoning markers	0.18	"prove", "theorem", "step by step"
Code presence	0.15	"function", "async", "import", "```"
Simple indicators	0.12	"what is", "define", "translate"
Multi-step patterns	0.12	"first...then", "step 1", numbered lists
Technical terms	0.10	"algorithm", "kubernetes", "distributed"
Token count	0.08	short (<50) vs long (>500) prompts
Creative markers	0.05	"story", "poem", "brainstorm"
Question complexity	0.05	Multiple question marks
Constraint count	0.04	"at most", "O(n)", "maximum"
Imperative verbs	0.03	"build", "create", "implement"
Output format	0.03	"json", "yaml", "schema"
Domain specificity	0.02	"quantum", "fpga", "genomics"
Reference complexity	0.02	"the docs", "the api", "above"
Negation complexity	0.01	"don't", "avoid", "without"

Weighted sum → sigmoid confidence calibration → tier selection.

Tier → Model Mapping

Tier	Primary Model	Cost/M	Savings vs Opus
SIMPLE	Gemini 2.5 Flash Lite	$0.10	99.8%
MEDIUM	Gemini 3 Flash	$0.50	90%
COMPLEX	Claude Sonnet 4.5	$3.00	40%
REASONING	Gemini 3 Pro	$2.00	60%

---

Models

16 models across 9 providers, one API key:

Model	Input $/M	Output $/M	Context	Reasoning
Google
gemini-3-flash-preview	$0.50	$3.00	1M	*
gemini-3-pro-preview	$2.00	$12.00	1M	*
gemini-2.5-flash-lite	$0.10	$0.40	1M	*
Anthropic
claude-sonnet-4.5	$3.00	$15.00	1M	*
claude-opus-4.6	$5.00	$25.00	200K	*
OpenAI
gpt-5.2	$1.75	$14.00	400K	*
gpt-5.1-codex	$1.25	$10.00	400K	*
gpt-5-nano	$0.05	$0.40	400K	*
xAI
grok-4.1-fast	$0.20	$0.50	2M	*
grok-4-fast	$0.20	$0.50	2M	*
grok-code-fast-1	$0.20	$1.50	256K	*
DeepSeek
deepseek-v3.2	$0.25	$0.38	164K	*
Moonshot
kimi-k2.5	$0.45	$2.50	262K	*
MiniMax
minimax-m2.1	$0.27	$0.95	197K	*
Z.AI
glm-4.7	$0.40	$1.50	203K	*
Arcee AI
trinity-large-preview (free)	$0.00	$0.00	131K

Full list: src/models.ts

---

Architecture

┌─────────────────────────────────────────────────────────────┐
│                     Your Application                         │
└─────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────┐
│                   ClawRouter (localhost:8402)                 │
│  ┌─────────────────┐  ┌─────────────────┐  ┌─────────────┐ │
│  │ Weighted Scorer │→ │ Model Selector  │→ │ Bearer Auth │ │
│  │  (14 dimensions)│  │ (cheapest tier) │  │  (API key)  │ │
│  └─────────────────┘  └─────────────────┘  └─────────────┘ │
└─────────────────────────────────────────────────────────────┘
                              │
                              ▼
┌─────────────────────────────────────────────────────────────┐
│                      OpenRouter API                          │
│  → OpenAI | Anthropic | Google | DeepSeek | xAI | Moonshot   │
└─────────────────────────────────────────────────────────────┘

Routing is client-side — open source and inspectable.

Source Structure

src/
├── index.ts          # Plugin entry point
├── provider.ts       # OpenClaw provider registration
├── proxy.ts          # Local HTTP proxy + Bearer auth
├── models.ts         # 16 model definitions with pricing
├── auth.ts           # API key resolution
├── logger.ts         # JSON usage logging
├── dedup.ts          # Response deduplication (prevents double-charge)
└── router/
    ├── index.ts      # route() entry point
    ├── rules.ts      # 14-dimension weighted scoring
    ├── selector.ts   # Tier → model selection
    ├── config.ts     # Default routing config
    └── types.ts      # TypeScript types

---

Configuration

Override Tier Models

# openclaw.yaml
plugins:
  - id: "clawrouter-openrouter"
    config:
      routing:
        tiers:
          COMPLEX:
            primary: "openai/gpt-5.2"
          SIMPLE:
            primary: "deepseek/deepseek-v3.2"

Override Scoring Weights

routing:
  scoring:
    reasoningKeywords: ["proof", "theorem", "formal verification"]
    codeKeywords: ["function", "class", "async", "await"]

---

Programmatic Usage

Use without OpenClaw:

import { startProxy } from "clawrouter-openrouter";

const proxy = await startProxy({
  apiKey: process.env.OPENROUTER_API_KEY!,
  onReady: (port) => console.log(`Proxy on port ${port}`),
  onRouted: (d) => console.log(`${d.model} saved ${(d.savings * 100).toFixed(0)}%`),
});

// Any OpenAI-compatible client works
const res = await fetch(`${proxy.baseUrl}/v1/chat/completions`, {
  method: "POST",
  headers: { "Content-Type": "application/json" },
  body: JSON.stringify({
    model: "clawrouter/auto",
    messages: [{ role: "user", content: "What is 2+2?" }],
  }),
});

await proxy.close();

Or use the router directly:

import { route, DEFAULT_ROUTING_CONFIG, ROUTER_MODELS } from "clawrouter-openrouter";

// Build pricing map
const modelPricing = new Map();
for (const m of ROUTER_MODELS) {
  modelPricing.set(m.id, { inputPrice: m.inputPrice, outputPrice: m.outputPrice });
}

const decision = route("Prove sqrt(2) is irrational", undefined, 4096, {
  config: DEFAULT_ROUTING_CONFIG,
  modelPricing,
});

console.log(decision);
// {
//   model: "openai/gpt-5.2",
//   tier: "REASONING",
//   confidence: 0.97,
//   method: "rules",
//   savings: 0.93,
//   costEstimate: 0.041,
// }

---

Performance Optimizations

• SSE heartbeat: Sends headers + heartbeat immediately, preventing upstream timeouts
• Response dedup: SHA-256 hash → 30s cache, prevents double-charge on retries

---

Troubleshooting

"Unknown model: clawrouter/auto"

This error means the ClawRouter plugin isn't loaded.

Fix:

# 1. Verify plugin is installed
openclaw plugins list

# 2. If not installed
openclaw plugins install clawrouter-openrouter

# 3. Restart OpenClaw after installing

# 4. Verify proxy is running
curl http://localhost:8402/health
# Should return {"status":"ok"}

"OpenRouter API key not found"

Fix: Set the environment variable:

export OPENROUTER_API_KEY=sk-or-...

Get your key at openrouter.ai/keys.

Port 8402 already in use

Fix:

# Find what's using the port
lsof -i :8402

# Kill it or restart OpenClaw

---

Development

git clone https://github.com/Not-Diamond/SimpleClawRouter.git
cd SimpleClawRouter
npm install
npm run build
npm run typecheck

---

License

MIT

---

Built on BlockRun's ClawRouter — powered by OpenRouter