Suanlimao
← Back to all tutorials

Connect your tools

SillyTavern

The roleplay front end: set the Chat Completion source to Custom (OpenAI-compatible).

About 5 minFor:People who run their own SillyTavern

Before you start

  • SillyTavern is up and running
  • A key and a model name

Step 1 · STEP 0x01

Choose Chat Completion and the custom source

API Connections (plug icon)
1

API

Chat Completion
2

Chat Completion Source

Custom (OpenAI-compatible)
Diagram only — the real screen in your version may differ

Click the plug icon to open the API connections panel, set the API type to Chat Completion and the source to Custom (OpenAI-compatible).

Create a key →

Step 2 · STEP 0x02

Enter the base URL and key

Custom (OpenAI-compatible)
1

Custom Endpoint (Base URL)

https://api.suanlimao.shop/v1
2

Custom API Key

<YOUR_API_KEY>
3
Connect
Diagram only — the real screen in your version may differ
  • Enter https://api.suanlimao.shop/v1 as the base URL. Official hints: add /v1 if you have connection trouble, and never add /chat/completions.
  • Click Connect.

Open the model plaza →

Step 3 · STEP 0x03

Pick a model and send a test message

Custom (OpenAI-compatible)
1

Model dropdown

<MODEL_NAME>
2

Enter a Model ID (if the list is empty)

<MODEL_NAME>
3
Test Message
Diagram only — the real screen in your version may differ
  • Our service implements /v1/models, so the dropdown fills after connecting; if it is empty, type the model ID by hand.
  • Click Test Message; a reply means it works.
Don't guess the model name: open the model plaza, click “Copy model id” on a model card and paste it. A wrong model name is the most common error.
The docs say compatibility with every endpoint is not guaranteed and custom-endpoint issues are unsupported. If SillyTavern keeps flagging the endpoint as down, tick “Bypass API status check”.

✓ How to tell it worked

  • The test message gets a reply, character chats work and the Usage page shows the requests.

FAQ

Connection fails.

Per the docs: make sure the address ends in /v1 without /chat/completions, then check the key.

Tool calling does not work.

Per the docs, tool calling is unsupported when a “no tools” post-processing option is selected; switch back to the default.

Can I use the Claude reverse proxy option?

The docs do not document that option, so we make no promises; use Custom (OpenAI-compatible) with a Claude-family model instead.

Related tutorials

Still stuck?

Read the troubleshooting tutorials first. If that does not help, open a ticket with the request_id, time, model and the exact error (never paste your key).