> Markdown twin of https://conduitllm.com/use-cases.html > What people use Conduit for — one free desktop AI client for your own API keys, local models, MCP tools, private work, your documents, and your own brand. --- 1. [Conduit](https://conduitllm.com/) 2. Use cases Use cases # One app. *Six reasons* to want it. Conduit is a free, open-source desktop client for every major AI model and the ones on your own machine. Here is what people actually use it for — and, on each page, what it doesn't do. For heavy AI users ### Pay per token, not per seat Your own API keys for every provider, in one app. No subscription and no markup on top of what the provider charges. [Bring your own key](https://conduitllm.com/use-cases/bring-your-own-key.html) For local-model users ### Local models, with a real app around them Ollama, LM Studio, or your own server — with documents, tools, and artifacts, and a cloud model one keystroke away when you want one. Run models locally For developers ### An MCP client that asks first Local stdio and remote servers with OAuth, one-click install from the official registry, and consent graded by what each tool can do. Use it as your MCP client For sensitive work ### An AI assistant that doesn't phone home No telemetry, no account, local storage. Pick a local model and switch on local-only mode, and it can run with no internet at all. Keep it private For research and analysis ### Ask your own files PDF, Word, CSV, Markdown, and text, parsed on your machine — with every answer pointing at the exact passage it used. Chat with your documents For AI and MCP product companies ### Put it under your own brand Your name, logo, and colours on a stock install today. Branded installers built from source for early design partners. White-label Conduit Quick pick ## If this sounds like you… | If you… | Start here | | --- | --- | | pay for two or three AI subscriptions and use each for a different model | [Bring your own key](https://conduitllm.com/use-cases/bring-your-own-key.html) — then check the [break-even calculator](https://conduitllm.com/api-vs-subscription.html) | | already run Ollama or LM Studio and want more than a bare chat box | [Local models](https://conduitllm.com/use-cases/local-llm.html) | | want MCP servers with a model other than one vendor's, or on Windows | [MCP client](https://conduitllm.com/use-cases/mcp-client.html) | | can't put client or patient material into a cloud chat account | [Private AI](https://conduitllm.com/use-cases/private-ai.html) | | need answers from a pile of PDFs, contracts, or spreadsheets — with sources | [Chat with your documents](https://conduitllm.com/use-cases/chat-with-documents.html) | | sell an MCP server or AI product and demo it inside someone else's app | [White-label](https://conduitllm.com/whitelabel.html) | ## Choosing a starting point **Do I have to pick one use case?** No. They are the same app. Most people start with one — usually their own API keys or a local model — and pick up the rest as they need it. A chat can use a cloud model, a local document collection, and an MCP server at the same time. **Is any of this a paid tier?** No. Every use case on this page is the same free, open-source (AGPL-3.0) app. You pay your model provider directly, or nothing at all with a local model. The one thing that is not generally available is packaged white-label installers, which are early access for design partners — see [the white-label page](https://conduitllm.com/whitelabel.html). **Can a team use Conduit?** Each person installs it and uses their own keys or a shared local server. There is no Conduit account, no admin console, and no sync between machines — conversations stay on the machine they happened on. If your team needs one shared, synced workspace, a hosted product is the better fit today. ## Try it on your own machine Free and open source. No account. Bring a key, or point it at a model you already run. [Download Conduit](https://conduitllm.com/#get) [See every feature](https://conduitllm.com/features.html) A release candidate: installers aren't OS code-signed yet, so expect a SmartScreen or Gatekeeper prompt on first launch.