modbender RelayPlane
Cut your AI API costs by 40-60% with intelligent model routing.
What It Does
RelayPlane is a local proxy that routes your LLM requests to the optimal model based on task complexity. Simple tasks go to cheaper models (Haiku), complex reasoning stays on premium models (Opus).
Installation
Install the proxy globally:
npm install -g @relayplane/proxy
Quick Start
# 1. Start the proxy
relayplane-proxy
# 2. Point OpenClaw at it (add to your shell config)
export ANTHROPIC_BASE_URL=http://localhost:3001
export OPENAI_BASE_URL=http://localhost:3001
# 3. Run OpenClaw normally - requests now route through RelayPlane
Commands
Once installed, use the CLI directly:
| Command | Description |
|---|---|
relayplane-proxy |
Start the proxy server |
relayplane-proxy stats |
View usage and cost breakdown |
relayplane-proxy telemetry off |
Disable telemetry |
relayplane-proxy telemetry status |
Check telemetry setting |
relayplane-proxy --help |
Show all options |
Configuration
The proxy runs on localhost:3001 by default. Configure via CLI flags:
relayplane-proxy --port 8080 # Custom port
relayplane-proxy --host 0.0.0.0 # Bind to all interfaces
relayplane-proxy --offline # No telemetry, no network except LLM APIs
relayplane-proxy --audit # Show telemetry payloads before sending
Environment Variables
Set your API keys before starting:
export ANTHROPIC_API_KEY=sk-ant-...
export OPENAI_API_KEY=sk-...
# Optional: Google, xAI
export GEMINI_API_KEY=...
export XAI_API_KEY=...
Privacy
- Your prompts stay local — never sent to RelayPlane
- Anonymous telemetry — only token counts, latency, model used
- Opt-out anytime —
relayplane-proxy telemetry off - Fully offline mode —
relayplane-proxy --offline