Enterprise LLM gateway — one API shape, aligned catalog, unified billing

One OpenAI-compatible endpoint; capabilities delivered through the products and contracts you purchase. The catalog shows what you can use—backend supplier topology stays private. Keys, usage, and billing in one place.

Get Started Browse models & prices

Try a request

routic — examples

curl -s https://api.routic.ai/v1/chat/completions \
  -H "Authorization: Bearer $ROUTIC_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o-mini",
    "messages": [{"role": "user", "content": "Hello from Routic"}]
  }'

What we offer

Familiar SDKs and endpoints

Integrate via an OpenAI-compatible path; choose entitled products with parameters such as model.

Catalog matches gateway configuration

The models shown on your site and in admin are the same ones your API calls can select.

Reduce gaps between what you tell customers and what integrations actually get; entitlements follow your contract and account setup.

Keys and usage you can govern

Budgets and access per key or team; usage and billing in one view.

Why Routic?

How catalog, gateway, and billing fit together

Unified access shape

Use an OpenAI-compatible client; pick a published SKU via parameters such as model. Each SKU has defined capabilities, pricing, and entitlements (subject to your account catalog).

Granular Control

You can set budgets, rate limits, and model access per key or per team.

Centralized Billing

One bill across products. Prepaid or postpaid options with transparent usage tracking.

B2B-grade books & ops

Usage, reconciliation, and billing for teams—aligned with published products and gateway config to reduce misconfiguration and churn risk.

Who typically starts here

Personas are for self-selection—not customer stories. Entitlements follow your agreement.

Builders and small teams

You already use an OpenAI-compatible client or your own stack and want to point it at the gateway and try entitled SKUs from the catalog.

Engineering leads

You need one integration shape, control over keys and model access, and sales/engineering alignment on what’s published.

Platform and finance ops

You care about billing, reconciliation, and fewer support tickets when catalog and gateway configuration drift.

How we compare

A quick framing vs common alternatives (illustrative—your contract prevails).

How we compare
Topic	Routic	Direct multi-vendor	Typical aggregator
Single OpenAI-compatible integration	Ready out of the box	Integrate and auth per vendor	Usually supported
Catalog matches contract pricing	Products aligned with gateway config	You maintain mappings and drift	Varies by platform and plan
No public backend supplier topology	Supplier details not marketed	You own the picture	Depends on disclosure policy
Unified billing / reconciliation	One bill and usage view	Fragmented bills, higher ops cost	Partial—depends on platform
Enterprise quotas, keys, and access	Keys, budgets, and model access	Build your own control plane	Common; depth varies

Service flow

Same capability layers as common AI gateways: your application, gateway governance and delivery, and model products on the right (icons are illustrative—not vendor logos).

Your AI APP

OpenAI-compatible client or your own stack

Traffic enters the gateway

Routic API Gateway

Unified auth, policy, and metering—productized outward.

OBSERVABILITYTraces, latency, errors, and usage visibility

GOVERNANCEKeys, quotas, access, and compliance policy

ROUTINGDeliver by product config—backend topology not shown publicly

PROMPT MANAGEMENTTemplates, versioning, and rollout

GUARDRAILSSafety and content guardrails

FINOPSBudgets, billing, reconciliation, and cost insight

Delivered by product policy

Broad model catalog

Published SKUs and entitled capabilities (per contract)

Custom Models Fine-Tuned Models

Layout matches the usual “app → gateway → models” story.

What a response looks like (illustrative)

A minimal chat.completion-style JSON example—real fields follow your environment and documentation.

example.json

{
  "id": "chatcmpl-placeholder",
  "object": "chat.completion",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello from Routic (example)."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 12,
    "completion_tokens": 8,
    "total_tokens": 20
  }
}

IDs and token counts are placeholders; actual responses come from your gateway.