AI models

Custom and self-hosted providers

Point Postout at any OpenAI-compatible endpoint, including services like OpenRouter or a model server you run yourself.

Updated

Alongside the built-in providers, Postout can draft with any service that speaks the OpenAI-compatible chat API. That includes hosted routers like OpenRouter and model servers you run yourself, such as Ollama, LM Studio, or llama.cpp.

Setting it up

  1. Open Settings → Models and add the Custom provider.
  2. Enter the server’s base URL, for example https://openrouter.ai/api/v1 or https://my-server.example.com/v1.
  3. Enter the model name to use.
  4. Add an API key if the service needs one. Most local servers don’t.

Both a base URL and a model name are required. If either is missing, Postout will tell you the custom provider isn’t fully configured when you try to create a draft.

Tips for self-hosted models

  • Use HTTPS. Addresses that start with https:// work most reliably. Plain http:// servers may not connect.
  • Pick a capable model. Drafting in a particular voice, within a character limit, for several platforms at once is a lot to ask. Small local models can struggle with longer templates. If results look off, try fewer platforms at a time or a larger model.
  • Reaching home from anywhere. A server on your home network isn’t reachable when you’re out. A private network like Tailscale connects your devices to it securely, and tailscale cert can give it an HTTPS certificate.

Privacy

Your idea, profile, and platform instructions go from your device straight to the address you entered, and nowhere else. See Privacy.

Didn't find your answer? Contact support. We read everything.