> ## Documentation Index > Fetch the complete documentation index at: https://dev-docs.altrum.ai/llms.txt > Use this file to discover all available pages before exploring further. # AI Gateway > Send your applications' and coding agents' model traffic through Altrum, so every request is governed. This is in development. Details may change before it is available. The AI Gateway is one address for every model call your organization makes. Your applications, SDKs and coding agents point at Altrum instead of at the model provider, and keep working as before. Altrum sends each request on to your provider, and on the way: * checks it against your [guardrails](/guardrails/at-runtime), before the model and on the answer, * checks the key, the models it may use, its [budget and rate limits](/ai-gateway/budgets-and-limits), * routes it to the right provider credential, and to a [fallback](/ai-gateway/models-and-routing) if the provider fails, * connects your [MCP tools](/ai-gateway/mcp), [prompts](/ai-gateway/prompts) and [cache](/ai-gateway/semantic-cache), * records what happened, with its cost, in the [request log](/ai-gateway/request-logs). ## Start without registering anything You can use the AI Gateway as soon as your organization is set up, before you register any AI system. Add your provider credentials, create a key, and send requests. Organization [guardrails](/guardrails/organization) govern that traffic. When you later register the AI system that traffic belongs to, attach the key to it, and the AI system's assessment and guardrails take over. ## Where you manage it | Place | What you manage there | | - | - | | **AI Gateway** in the sidebar | Provider credentials, organization and personal keys, organization budget and rate limit, model access and routing, organization guardrails, MCP servers, prompts, cache, content logging, OpenTelemetry export, and the request log for the whole organization. | | An AI system's **AI Gateway** tab | That AI system's keys, budget and rate limit, models, routing, cache and logging choices, MCP tools, and its own request log. | ## What it supports | Area | Supported | | - | - | | Request formats | OpenAI chat completions and Responses, Anthropic messages, Google Gemini. See [Endpoints](/ai-gateway/endpoints). | | Model providers | OpenAI, Anthropic, Bedrock, Google Gemini, Google Vertex AI and Azure (Azure OpenAI and Azure AI Foundry), with your own credentials. See [Provider credentials](/ai-gateway/provider-credentials). | | Clients | The OpenAI, Anthropic and Google SDKs, Claude Code, Codex CLI, Gemini CLI, Cursor and Claude Desktop. See [SDKs](/ai-gateway/sdks) and [Coding agents](/ai-gateway/coding-agents). | | Streaming | Yes, for every request format. | ## Who can do what Owners and admins manage the AI Gateway. When an admin allows it, any member can create a personal key for their own coding agent. Every member can see the request log, without request content. See [Virtual keys](/ai-gateway/virtual-keys). This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.