winsorllc Model Router
Configure routing between LLM providers and models with fallback support.
Features
- config: Set up routing rules per task type
- list: Show available providers and models
- route: Route requests to appropriate model
- fallback: Configure fallback providers
- weights: Set model usage weights
- status: Check provider health
Usage
# List available providers
./scripts/model-router.js --command list
# Configure routing
./scripts/model-router.js --command config --task coding --provider anthropic --model claude-sonnet-4
# Set fallback
./scripts/model-router.js --command fallback --primary anthropic --fallback openai,groq
# Check status
./scripts/model-router.js --command status
# Route a request
./scripts/model-router.js --command route --task research --prompt "Explain quantum computing"
Routing Rules
Routing can be based on:
- Task type (coding, research, creative, reasoning)
- Token count
- Model capability
- Provider availability
- Cost optimization
Examples
| Task | Command |
|---|---|
| List providers | model-router.js --command list |
| Config coding | model-router.js --config --task coding --provider anthropic --model opus |
| Set fallback | model-router.js --fallback --provider anthropic --fallback openai,groq |
| Check status | model-router.js --status |
| Route request | model-router.js --route --task summarize --file article.txt |
Notes
- Supports major providers: OpenAI, Anthropic, Groq, etc.
- Automatic failover to fallback providers
- Cost and quality optimized routing
- Health checks on providers