One API. Every AI model.

OpenAI-compatible AI gateway (LLM gateway) and OpenRouter alternative. Switch between models with one parameter. Exfiltration Guard by default. Unified billing. Hosted in Québec.

Anthropic OpenAI Google Cohere Mistral +12
curl · chat.completions 200 OK
curl https://router.signalorange.ca/v1/chat/completions \
  -H "Authorization: Bearer $MERIDIAN_KEY" \
  -d '{
    "model": "anthropic/claude-sonnet-4.7",
    "messages": [{ "role": "user", "content": "Hello" }]
  }'
Guard: OK PII redacted · 0 stored ↳ 240ms · Montréal

The right model for every task.

Every AI model has its strengths. Router gives you all of them, without having to manage integrations.

01

One catalog, every model

Vision, code, reasoning, translation. OpenAI, Anthropic, Google, Cohere, Mistral.

02

Switch models in one parameter

A better model ships? Change the call. No refactor, no migration.

03

Intelligent routing

Simple tasks to a fast, affordable model; complex tasks to a premium one. Cost and latency optimized.

Complete API, OpenAI-compatible.

Drop-in replacement. Same request format, every provider behind a single endpoint.

Chat Completions

Streaming, JSON mode, system prompts. Compatible with every OpenAI SDK.

Vision

Send images to models that support them. Same multimodal format.

Function Calling

Unified tool use and function calling, regardless of the provider behind it.

Embeddings

Generate embeddings for your data through the same endpoint.

Automatic fallback

A provider goes down? The request automatically falls back to an equivalent model.

Model catalog

Endpoint /models to list available models, their capabilities, pricing, and limits.

Guard adds only 11ms to the initial call.

request lifecycle · req_9k2abc ✓ 200 OK · 838ms
00:00.002 ingress
2ms
00:00.010 guard.scan
8ms
00:00.011 redact
3ms
00:00.013 route
1ms
00:00.013 model
→ anthropic/claude-sonnet-4
820ms
00:00.833 response
4ms
Total latency 838ms · p95 under 1s ✓ Guard overhead: 11ms (1.3%)

Every request is traced end-to-end. Guard adds ~11ms to an average response.

Your data never leaves unprotected.

Router is the only Canadian AI gateway with a built-in technical safeguard against sensitive data exfiltration.

Active protection, not a policy

Every request is scanned before reaching the provider. PII, credit cards, tokens, secret keys, .env files — detected and redacted automatically.

No data stored

Router keeps no request or response data. We only select providers with the same contractual guarantee.

Hosted in Québec, PIPEDA + Loi 25 compliant

Québec infrastructure. Your data is processed under Québec jurisdiction, all the way to the AI provider.

APPLICATION web · mobile · backend HTTPS ROUTER 🛡 Guard redact: pii · keys · .env · cc → Model selection fallback · latency · cost OpenAI Anthropic Google Mistral + 14 additional models response stream OPENAI-COMPATIBLE · ONE ENDPOINT

One invoice. No surprises.

No more credit packs or juggling multiple provider accounts.

Without Router

  • One account per provider
  • One credit card per provider
  • Credit packs to manually reload
  • 5 different invoices to reconcile
  • No consolidated usage visibility

With Router

  • One account, one (or several) API keys
  • Post-monthly billing at real cost
  • Transparent fixed margin, no hidden markup
  • One unified invoice
  • Budget caps and limits per organization

Enterprise billing · Net 30 · No minimum commitment

Every model. One endpoint.

Live metrics on every request. Unified billing across all providers. One OpenAI-compatible endpoint.

router / live live
Requests
2,481/min
↑ 12%
Guard blocks
147
PII · keys · .env
p50 latency
342ms
↓ 18ms
Traffic by provider
OpenAI
42%
Anthropic
28%
Google
20%
Mistral
10%
Full catalog
Anthropic OpenAI Google Cohere Mistral Meta xAI Perplexity Groq Together AI Fireworks DeepSeek Qwen +5

Figures illustrative of a production deployment. Live metrics available on the dashboard.

Ready to simplify your AI infrastructure?

Join the private beta. We'll reach out for a tailored demo.

No commitment · Response within 24 hours