Overview
If a service says it is “OpenAI-compatible”, Loom can talk to it. You give it an address, say how it wants the key, and add any extra headers it needs.
Loom does not support signing in with a ChatGPT, claude.ai or Gemini app subscription, or GitHub Copilot, because those products offer no sign-in for other apps.
Before you start
- 1Find the service’s base URL in its docs. It usually ends in
/v1and is what you would put in an OpenAI client. - 2Get an API key from the service, if it uses one.
- 3In Loom choose Custom, then fill Endpoint. Open Advanced to set how the key is sent: Bearer (the usual), Key header with the header’s name, or None.
- 4Add any custom headers the service wants, and any extra request fields as JSON.
Cost. Whatever the service charges.
Add it to Loom
- 1Open Loom, then Providers in the menu, and choose Add provider.
- 2Pick Custom (the search box finds it).
- 3Fill in the fields in the table below.
- 4Choose Connect. Loom checks the key and says how many models it found. If it cannot check, it says so, and you can Save anyway and try a message.
- 5Open the model chip below the message box and pick a model.
- Use Test connection. If the service has a model list at
/models, Loom shows it. If not, type a model name in Default model.
| Field in Loom | What to put | Needed |
|---|---|---|
| API key | Only if the server or service asks for one | Optional |
| Endpoint | you enter it | Yes |
| Name | Anything you like. It is shown in the model picker | Optional |
At a glance
| Appears in Loom as | Custom |
| Sign-in | Your choice: Bearer, a named key header, or none |
| Models | Loom fetches the list from your account |
| Tools (connectors, search) | Depends on the model |
| Photos | Depends on the model |
| PDFs | Sent as an OpenAI file part |
Addresses you may need
These providers already have their own entry in Loom, so you normally do not need Custom for them. They show what a base URL looks like.
| Service | Base URL |
|---|---|
| Ollama Cloud | https://ollama.com/v1 |
| LM Studio | http://localhost:1234/v1 |
| vLLM or llama.cpp | http://localhost:8000/v1 |
| Perplexity | https://api.perplexity.ai |
| Groq | https://api.groq.com/openai/v1 |
| Together AI | https://api.together.ai/v1 |
| Fireworks AI | https://api.fireworks.ai/inference/v1 |
| Cerebras | https://api.cerebras.ai/v1 |
| SambaNova | https://api.sambanova.ai/v1 |
| OpenRouter | https://openrouter.ai/api/v1 |
| Mistral | https://api.mistral.ai/v1 |
| DeepSeek | https://api.deepseek.com |
| xAI Grok | https://api.x.ai/v1 |
If something goes wrong
| What you see | What to do |
|---|---|
| 404 | The address is probably missing or has an extra /v1. Compare with the service’s docs. |
| 401 | Wrong key, or the service wants a different header. Try Advanced → Key header. |
| A service that is not on the list | If it offers an Anthropic-style or OpenAI-style address, Custom usually works. |