Providers
Loach doesn't ship with built-in models. It connects to a provider you choose — local or remote. Both kinds are available side-by-side; switch from the chat header.
Ollama (local)
Ollama is the easiest way to run open-weight models on your machine. Loach probes http://localhost:11434 on launch; any models you've pulled appear in the chat header automatically.
Setup
- Install Ollama from ollama.com/download.
- Start the server:
ollama serve(or just open the Ollama app — it runs in the background). - Pull at least one model:
ollama pull gemma3:4b. - Open Loach. The model appears in the chat-header dropdown.
The Models library
The Models tab inside Loach is a full UI on top of the Ollama HTTP API. From there you can:
- Pull new tags by name.
- Copy a model under a new name.
- Customize an existing model with a Modelfile (system prompt, parameters, templates).
- Delete models you no longer need.
Pointing at a remote Ollama
If Ollama is running on another machine on your network, change the base URL in Settings → Providers → Ollama. Make sure the remote server is reachable from your machine and listening on a routable interface.
OpenAI-compatible endpoints
Anything that speaks the OpenAI Chat Completions protocol works. Configure it in Settings → Providers:
- Paste the API key (stored in your OS credential manager — never in plain text on disk).
- Set the base URL for the endpoint.
- Save. The provider becomes selectable from the chat header.
Common base URLs
| Provider | Base URL |
|---|---|
| OpenAI | https://api.openai.com/v1 (default) |
| vLLM | http://localhost:8000/v1 |
| LM Studio | http://localhost:1234/v1 |
| LiteLLM | http://localhost:4000/v1 |
llama.cpp llama-server | http://localhost:8080/v1 |
| Groq | https://api.groq.com/openai/v1 |
| OpenRouter | https://openrouter.ai/api/v1 |
| Together | https://api.together.xyz/v1 |
| Fireworks | https://api.fireworks.ai/inference/v1 |
| Mistral | https://api.mistral.ai/v1 |
Only one OpenAI-compatible endpoint is active at a time. To switch providers (say, from Groq to OpenRouter) change the base URL and key in Settings — your chats with the previous endpoint are preserved.
Default model
Pick the model new chats should open with — independently for each provider. New conversations honour the default; you can still change the model per-chat from the header.
Preloading and Low VRAM
Two settings worth knowing about for local providers:
- Model preloading — optionally warm your default local model into VRAM at launch, so the first message streams faster.
- Low VRAM mode — a global or per-chat toggle that sends Ollama's
low_vramflag with every request. Useful on lower-spec devices.