OPEN BETA

The AI router that keeps your
private data private

SmartRouter automatically routes your AI requests — sensitive queries stay local, everything else goes to the fastest free model.

Request API key Read the docs ↓
🔒

Auto-privacy routing

PII detection runs locally. Sensitive messages never leave your infrastructure.

⚡

15+ models, one API

Groq, Mistral, Gemini, Cerebras, Claude, local Ollama — one endpoint, best model selected automatically.

💰

Mostly free routing

80%+ of requests route to free-tier models. You only pay for the API key, not per-token.

🧠

RAG-first

Checks your knowledge base before hitting any LLM. Repeated questions cost nothing.

Pricing

Free

$0
forever
  • 100 requests / month
  • All models
  • Privacy routing
Get free key

Hobby

$5
per month
  • 1,000 requests / month
  • All models
  • Privacy routing
  • Usage stats endpoint
Get started — $5/mo

Pro

$20
per month
  • 10,000 requests / month
  • All models + local Ollama
  • Strict privacy mode
  • Priority routing
Get started — $20/mo

Quick start

Send a POST request with your API key in the X-Api-Key header:

curl -X POST https://smartrouter.connectionlynx.info/route \ -H "X-Api-Key: sk_izi_YOUR_KEY" \ -H "Content-Type: application/json" \ -d '{"message": "Explain quantum entanglement simply"}'

Response:

{ "text": "Quantum entanglement is when two particles...", "agent_used": "groq", "model": "llama-3.1-70b", "task_type": "research", "privacy_level": 1, "cost_usd": 0.0, "success": true }

FAQ

What makes this different from just calling Groq/Gemini directly?
SmartRouter classifies your query first. If it detects personal data, it routes to a local model instead of any cloud API. You don't have to think about privacy — it's automatic.
What happens when I hit my monthly limit?
You get a 429 response with an upgrade prompt. No surprise charges.
Can I force a specific model?
Yes — pass "force_agent": "groq" in your request to bypass automatic routing.
Is there a streaming endpoint?
Not yet. It's on the roadmap.