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

# LLMGrid Overview

> A unified platform for routing, governing, and observing LLM usage across models, tools, and agents.

## What is LLMGrid?

**LLMGrid** is an enterprise‑ready AI gateway and orchestration platform that provides a **single control plane** for using large language models, tools, and agents. It centralizes access, governance, routing, safety, and observability—without requiring application rewrites.

LLMGrid exposes an **OpenAI‑compatible API**, so existing SDKs and frameworks work by simply pointing to the LLMGrid proxy.

<Frame>
  <img src="https://mintcdn.com/llmgrid/Pu--kCHp7HOHFrpj/images/image.png?fit=max&auto=format&n=Pu--kCHp7HOHFrpj&q=85&s=2652bbe4b8ef48a09e0389c96113199f" alt="Image" width="655" height="357" data-path="images/image.png" />
</Frame>

***

## Core Capabilities

### OpenAI‑Compatible Proxy

* Drop‑in replacement for OpenAI SDKs
* Minimal code changes (update `base_url` and API key)
* Supports chat, embeddings, streaming, tools, and function calling

***

### Model & Traffic Routing

* Route requests across models and providers
* Configure fallbacks and retry strategies
* Use aliases to keep application code stable during model changes
* Optimize for availability, latency, or efficiency

***

### Governance & Safety

* **Virtual Keys** for scoped access and limits
* **Guardrails** for input/output and tool enforcement
* **Budgets** and rate limits to prevent overuse
* **Tags** for routing, attribution, and segmentation

***

### Agents, Tools & Retrieval

* Register **Agents** with skills and capabilities
* Attach **Search Tools** for live retrieval and grounding
* Manage **Vector Stores** for RAG workflows
* Secure tool execution with pre‑execution checks

***

### Observability & Analytics

* Request and audit logs
* Usage and cost analytics by model, key, team, tag, or agent
* Cache analytics and health checks
* End‑to‑end visibility for debugging and audits

***

### Performance & Efficiency

* Response **Caching** (Redis‑backed) to reduce latency and repeat calls
* Semantic caching for similarity‑based reuse
* Centralized cost tracking and discounts

***

## How LLMGrid Fits In

LLMGrid sits between your applications and AI capabilities: Your App ↓ LLMGrid (Auth • Routing • Guardrails • Observability) ↓ Models • Tools • Vector Stores • Search

This architecture lets teams evolve models and controls independently of application code.

***

## Who Should Use LLMGrid?

* **Platform teams** needing centralized governance
* **Developers** shipping AI features quickly
* **Security & compliance** teams enforcing policies
* **FinOps** teams monitoring usage and cost
* **Enterprises** running multi‑model, multi‑tool AI workloads

***

## Getting Started

1. Create a **Virtual Key**
2. Point your OpenAI SDK to the LLMGrid proxy
3. Configure models and routing
4. Add guardrails, budgets, and observability
5. Iterate safely as usage grows

***

## Related Sections

* **API Reference** – OpenAI‑compatible endpoints
* **Models** – Configure available models
* **Router Settings** – Control routing and fallbacks
* **Guardrails** – Enforce safety and compliance
* **Usage & Logs** – Observe and analyze traffic
* **Security & Compliance** – Enterprise controls and governance