Skip to content
Build Perch
Browse documentation

Connect your first model

Short answer: Build Perch connects to three providers: OpenRouter (a paid gateway to hundreds of cloud models, needs an API key), Ollama (free local models, no key, but a separate install), and Ollama Cloud (your ollama.com subscription’s hosted models, needs a key). Open the Connect AI Model dialog and pick whichever fits how you want to work.

Which provider should I pick?

  • OpenRouter (remote). One API key gives you the live catalog (GPT, Claude, Gemini, Llama, and hundreds more across every major lab), with new models appearing the day OpenRouter lists them, no app update needed. You pay OpenRouter per message at the rate shown next to each model; it needs an internet connection. Pick this for the widest model choice.
  • Ollama (local, keyless). Free and fully offline: models run on your own hardware and nothing leaves your machine. It needs the separate Ollama application installed; Build Perch detects it automatically and manages the background service, but doesn’t install Ollama itself. Model quality and speed depend on your computer, especially available memory. Pick this for zero cost and full privacy.
  • Ollama Cloud (remote, keyed). Your own ollama.com subscription’s hosted models (larger open models running on Ollama’s servers instead of yours), a flat-rate way to use bigger models without a local GPU. Needs an ollama.com API key. Pick this if you want bigger open models than your hardware can run.

Media generation (images, speech, music, video) works only through OpenRouter; Ollama and Ollama Cloud are chat-only.

How do I connect my first model?

  1. With a workspace open, show the Chats panel (Ctrl+2) and click Add Model, the button that appears once you have a workspace but no chat yet.
  2. In the Connect AI Model dialog, pick a provider from the dropdown: OpenRouter, Ollama (Cloud), or Ollama (Local).
  3. For OpenRouter or Ollama Cloud without a saved key: the dialog’s Integrations window link opens it. There, select the provider under LLM Providers, click Get API key to open its key page in your browser (openrouter.ai/keys or ollama.com/settings/keys), copy a key, paste it into the provider’s row, and click Save; Build Perch tests the connection and confirms.
  4. For Ollama (Local): Build Perch checks whether Ollama is installed. If it isn’t, click Download Ollama, install it, then reopen this dialog. Once installed, Build Perch starts the Ollama server in the background and lists your installed models; if the list is empty, pull a model in Ollama first.
  5. Pick a model and click Connect. This creates your first chat.

Good to know

  • API keys live in the Windows Credential Manager, not in app files. Once saved, a key is never shown back to you; the field shows only a “Saved” status, so you replace a key rather than view it.
  • You can connect more than one model at a time (up to 20 active chats per workspace) and switch a chat’s model any time; every reply is stamped with the exact model that produced it.
  • Local Ollama models need no key, but they do need a capable machine; the picker shows each installed model’s size, quantization, and context window before you choose one.