> ## Documentation Index
> Fetch the complete documentation index at: https://gateway.consus.io/llms.txt
> Use this file to discover all available pages before exploring further.

# Overview

> Compliant AI infrastructure for companies handling sensitive data.

Consus Gateway is an API gateway that routes AI requests through FedRAMP High, DoD IL2, and DoD IL5 authorized government cloud environments. It's built for defense contractors and companies handling sensitive data who need access to frontier AI models without months of cloud integration and compliance paperwork.

## Who It's For

Consus Gateway is for companies who handle sensitive information but refuse to let compliance slow down their AI adoption.

## What You Get

* **Frontier AI models:** Claude, Gemini, and GPT models served from AWS GovCloud, GCP Assured Workloads, and Azure Government
* **Compliance by default:** Every request routes through FedRAMP High, DoD IL2, or DoD IL5 authorized infrastructure. No configuration required.
* **Zero-persistence architecture:** Prompts and responses exist only in ephemeral compute memory. Only billing metadata is stored.
* **OpenAI SDK compatibility:** Works with the OpenAI SDK, LangChain, LlamaIndex, Cline, OpenCode, and any tool that speaks the OpenAI API format
* **Vendor coding agents:** [Claude Code](/integrations/claude-code) targets `/v1/messages` for Claude; [Codex CLI](/integrations/codex) targets `/v1/responses` for GPT
* **Audit-ready logging:** 365-day API access logs, 3-year GCP audit retention, all KMS-encrypted. No prompt content in logs.

## How It Works

You call the standard OpenAI API: same SDK, same format, same tools. Consus Gateway authenticates your request, selects the correct government cloud backend based on the model and compliance level you specify, and returns the response. Your prompts never touch commercial cloud infrastructure and are never persisted.

## AI Model Security Compliance Distinction

AI models don't hold their own security authorization. The compliance boundary comes from the cloud infrastructure they run on. Every model on Consus Gateway is served through a cloud provider and deployment path that has been explicitly authorized. This means the model itself was in-scope for the cloud provider's FedRAMP or DoD authorization — not merely running on an authorized platform.

We track the (model × provider × impact level) matrix so you don't have to. See the [model catalog](/models) for the full list of available models and their authorization levels.

## Get Started

<CardGroup cols={2}>
  <Card title="Request access" icon="shield-check" href="https://tally.so/r/QKBYVg">
    Get an API key — typically approved within 24 hours.
  </Card>

  <Card title="Quickstart" icon="code" href="/getting-started">
    Two lines of config. Five minutes to your first request.
  </Card>
</CardGroup>

## *Coming Soon*

***Consus Code** is our AI coding agent built for engineers who work with sensitive data. It runs locally on your machine and leverages Consus Gateway for all LLM calls. Built using NIST 800.53 as our guide, Consus Code enforces guardrails that constrain what the agent can do and where data can flow, giving engineers a capable coding assistant without compromising the security posture of their environment.*

<CardGroup cols={3}>
  <Card title="Request early access" icon="rocket" href="https://tally.so/r/vGOdO8" />
</CardGroup>