AI Model Router API
Route prompts to the right LLM — cheapest model that fits your task. GPT-4o · Claude 3 · Gemini 1.5/2.0. No external API calls.
The Problem
Every LLM call uses the same model even when a cheap model would do. A classification task costs 100x more than necessary when sent to GPT-4. This API picks the right model automatically.
GET /route
Describe your task. Get back the optimal model, estimated cost, and reasoning.
GET /compare
Side-by-side cost breakdown across all 9 models for your token count.
GET /models
Full catalog: pricing, context windows, strengths per model.
POST /mcp
MCP-compatible — plug into any agent that supports JSON-RPC 2.0 tools.
Route a coding task to the cheapest capable model:
curl "https://api.lazy-mac.com/ai-model-router/route?prompt=write+a+quicksort+in+python&prefer_tier=economy"
# Response:
{
"selected_model": "gemini-1.5-flash",
"provider": "google",
"tier": "economy",
"task_classification": "code",
"estimated_cost_usd": 0.00003,
"reasoning": "Task classified as \"code\". gemini-1.5-flash has native strength in code tasks. Economy-tier model balances quality and cost."
}
Compare all models for a 2000-token analysis job:
curl "https://api.lazy-mac.com/ai-model-router/compare?task_type=analysis&input_tokens=2000"
Free
$0/mo
- 100 requests/day
- All routing endpoints
- Full model catalog
- MCP support
Pro
$29/mo
- Unlimited requests
- Priority routing
- Batch compare endpoint
- Priority support
Get API Key — $29/mo