Skip to content

ModelPort

Artificial Intelligence

One port for every AI model.

ModelPort is the gateway between your applications and the models they call. Connect a provider once, hand each team its own key, and control routing, budgets and usage from a single place instead of inside every codebase.

The problem

AI adoption outruns control

Teams start with one provider key pasted into one application. A few months later there are several providers, keys in many repositories, and no shared answer to who is spending what, on which model, for which product.

Changing a model means changing code. A provider outage becomes an application outage. Finance sees one invoice with no breakdown, and security cannot say where prompts are going.

  • Provider keys copied into applications and scripts
  • No per-team or per-application view of usage and cost
  • Switching models or providers requires code changes
  • One provider outage takes the feature down with it

Key capabilities

What ModelPort does

Put every large language model your organization uses behind one OpenAI-compatible endpoint. ModelPort issues keys per team and application, routes requests between providers and self-hosted models, and shows exactly what is being used and what it costs.

  • Unified API

    A single OpenAI-compatible endpoint in front of many providers and models, so existing SDKs and tools keep working.

  • Routing & fallback

    Send each request to the right model by rule, and fail over to an alternative when a provider is slow or unavailable.

  • Virtual keys

    Issue, scope and revoke keys per team, application or environment without ever sharing the provider credential.

  • Usage metering

    Requests, tokens and cost recorded per key, model and project, ready for chargeback and capacity planning.

  • Budgets & rate limits

    Spending caps and request limits per key, so an experiment cannot become a surprise invoice.

  • Request logging & audit

    A searchable record of which key called which model and when, with configurable retention.

  • Self-hosted models

    Register models running on your own servers next to hosted providers and expose both through the same endpoint.

  • Provider key vault

    Provider credentials are stored encrypted in one place and never reach application code.

How it works

From setup to steady-state operation

  1. 01

    Connect providers

    Add hosted providers and self-hosted model endpoints once.

  2. 02

    Issue keys

    Create virtual keys for each team, application and environment.

  3. 03

    Point your apps

    Change the base URL; existing OpenAI-compatible clients keep working.

  4. 04

    Set policy

    Define routing, fallbacks, budgets and rate limits per key.

  5. 05

    Observe

    Track usage and cost by team, model and project.

Architecture

Where ModelPort sits in your stack

ModelPort sits between your applications and every model they call.

  1. Your applications

    • Web & mobile apps
    • Internal tools
    • Agents & automations
  2. ModelPort gateway

    • Unified API
    • Routing & fallback
    • Virtual keys
    • Budgets & limits
  3. Governance

    • Usage metering
    • Request log
    • Provider key vault
  4. Models

    • Hosted LLM providers
    • Self-hosted models
    • Embedding models
Highlighted layers are provided by ModelPort. Other layers are your existing systems or other E-Net Communication products.

Operational benefits

What changes for your team

  • Cost you can explain

    Every token is attributed to a team, application and model.

  • Change models without redeploying

    Routing lives in the gateway, so swapping a model is a configuration change.

  • Fewer AI outages

    Fallback routes keep features working when one provider has a bad day.

  • Control over where data goes

    Decide which workloads may use hosted providers and which must stay on your own models.

Who it is for

Target customers

  • Software teams

    Product teams shipping AI features across several applications.

  • Enterprises

    IT and platform teams that need one governed entry point for AI usage.

  • Regulated organizations

    Banks and public bodies that must control and record where prompts are sent.

  • Service providers

    Hosting and telecom providers offering AI access to their own customers.

Deployment

Run it where you need it

Run the gateway inside your own network when prompts and logs must not leave your infrastructure.

  • Cloud

    Operated by E-Net Communication. No infrastructure to manage.

  • Private cloud

    Dedicated environment in your cloud account or ours.

  • On-premise

    Installed in your own data center, under your control.

  • Hybrid

    Edge components on-site with central services elsewhere.

Get started

See ModelPort in your environment.

Book a walkthrough with the E-Net Communication team. We will map ModelPort to your current systems and deployment requirements.