Configure in MiniTavern

App path: Settings → Models → Model link.

Create a link

1. Open MiniTavern → bottom Settings.
2. Models → Model link.
3. Tap New (top-right).
4. Model platform: Custom (OpenAI-compatible / custom Base URL — exact label follows the current App).
5. Fill the table for your tool, save; if Test is available, Test before setting as default.

Examples by tool

Replace YOUR_PC_IP with the PC LAN IP (never localhost).

Tool	Base URL example	Model name	API Key
Ollama	http://YOUR_PC_IP:11434/v1	Name from ollama list	Empty or ollama
LM Studio	http://YOUR_PC_IP:1234/v1	id from Local Server	Empty
KoboldCPP	http://YOUR_PC_IP:5001/v1	id from /v1/models	Empty
llama.cpp	http://YOUR_PC_IP:8080/v1	id from server	Empty
LocalAI	http://YOUR_PC_IP:8080/v1	LocalAI config name	Per auth settings
Jan	Address on the app API page	id from the list	Per app settings
text-generation-webui	http://YOUR_PC_IP:port/v1	id from the extension	Per extension settings

Aliases are up to you, e.g. Home Ollama, Office Kobold.

Set default and chat

1. In the model link list, set as Default.
2. Import or pick a character card and send a message.
3. The PC tool should show request activity; the App should stream or show a full reply.

Per-chat override (optional)

To temporarily use another local model for one chat: override the global default model link / parameters in chat settings.

Notes

• Do not put http://localhost:... or http://127.0.0.1:... on the phone.
• Prefer a Base URL ending in /v1, not /chat/completions.
• PC sleep, lid close, or network changes can change the IP.
• Ports are whatever you actually opened; the table is common defaults only.

Cannot connect? Troubleshoot. Pick a tool: Tools overview.