Search docs...

Start typing to search documentation

RisuAI

RisuAI Integration Guide

Connect UnoRouter to RisuAI through the Custom Models option in Advanced Settings.

What is RisuAI?

RisuAI is a local-first roleplay chat frontend that runs in the browser or as a desktop app. It supports any OpenAI-compatible endpoint, with configuration stored in browser local storage.

The official RisuAI wiki is behind the current UI. Steps below reflect RisuAI as of 2026-04.

Overview

RisuAI's Custom Models slot lets you paste a Request URL, API key, and model ID. The gotcha is a toggle called Autofill Request URL: when it is on, RisuAI rewrites the URL you paste, which silently breaks the connection.

Uncheck Autofill Request URL before pasting

If Autofill Request URL is on, RisuAI rewrites whatever URL you paste into its own default format, which breaks the connection with no error. Uncheck this toggle BEFORE you paste the Request URL. This is the single most common reason a RisuAI + custom proxy setup fails.

Quick Config

Copy these three values into the Custom Models fields in RisuAI's Advanced Settings.

text
Request URL: https://api.unorouter.ai/v1/chat/completions
API Key:    YOUR_API_KEY
Model ID:   claude-opus-4-5

Sign in to auto-fill your API key

Paste the full URL ending in `/v1/chat/completions`. RisuAI has no model dropdown in Custom Models mode; the Model ID is a manual text field.

Step-by-step setup

  1. 1
    Open Advanced Settings

    In RisuAI, open Settings, then Advanced Settings.

  2. 2
    Pick a model slot

    Click the Chat Bot model selector, scroll to the bottom of the model list, and choose Custom Models.

  3. 3
    Uncheck Autofill Request URL

    This toggle must be OFF before you paste anything. See the critical warning above.

  4. 4
    Paste the Request URL

    Paste the full URL ending in `/v1/chat/completions`.

  5. 5
    Paste the API key

    Paste your UnoRouter API key into the API Key field. Some RisuAI builds label this field Authorization; use that field if the API Key field does not appear.

  6. 6
    Paste the Model ID

    Type or paste the exact model name from UnoRouter's models page.

  7. 7
    Save

    Close the settings panel; RisuAI writes configuration to local storage.

  8. 8
    Send a test message

    Open any chat and send a test message to verify the response.

Why the autofill toggle matters

Autofill Request URL was added to prevent typos against named preset providers. For Custom Models, it becomes a foot-gun: RisuAI rewrites your custom URL to a default that does not point at us. If a RisuAI update flips this toggle back on, redo step 3 before debugging anything else.

Configure the Auxiliary Model

RisuAI uses a second model slot for translation, summary, and name-guessing tasks. Repeat steps 2 through 7 for the Auxiliary Model selector, using the same UnoRouter URL and key.

Troubleshooting

  • 404 or 'Invalid URL'. Autofill Request URL was on when you pasted; RisuAI rewrote the URL. Uncheck the toggle and re-paste the Request URL.
  • Worked before, no response now. A RisuAI update can re-enable Autofill Request URL silently. Open Custom Models and confirm it is still off.
  • 401 Unauthorized. RisuAI builds sometimes expose both API Key and Authorization fields; only one is read depending on version. Move your key to the other field and retry.
  • Model not found. The Model ID must match a model name exactly. Copy-paste from UnoRouter's models page; do not retype.
  • Translation or summary feature is broken. The Auxiliary Model is unconfigured. Repeat setup for the Auxiliary Model slot.
  • Streaming output is chunked or missing pieces. RisuAI's SSE parser expects plain `data:` events. A corporate proxy stripping the prefix will produce garbled streaming. Test on a non-corporate network if available.
  • Settings reset after browser restart. RisuAI stores configuration in browser local storage, which some privacy modes clear. Export your config to a file from Settings for backup.

Ready to chat?

Get your UnoRouter API key and run RisuAI against every model we support.