Integrations

Open WebUI → AI24X

Open WebUI is a popular self-hosted chat UI. Add AI24X as an OpenAI-compatible connection—one key for flash/pro and VIP named models, PayPal USD top-up.

1) Prerequisites

Sign up → Console → create an API key → keep a small credit balance. Open WebUI running (Docker or local).

2) Admin UI connection

  1. Sign in as admin → Settings → Connections (or Admin → Settings → Connections).
  2. OpenAI / OpenAI-compatible → add or edit connection.
  3. API Base URL: https://api.ai24x.com/v1 (stop at /v1).
  4. API Key: your AI24X key from Console.
  5. Save → refresh model list → enable flash (optional pro / ultra).
  6. New chat → select flash → send hello.

3) Docker / env alternative

Some installs prefer environment variables (names vary by Open WebUI version):

OPENAI_API_BASE_URL=https://api.ai24x.com/v1 OPENAI_API_KEY=YOUR_AI24X_API_KEY

If your image uses OPENAI_API_BASES / multi-connection JSON, keep the same base URL ending at /v1 and the same key.

4) Pitfalls

  • Base URL with /chat/completions appended → 405 (the UI adds the path).
  • 401 = bad, truncated, or rotated key—paste the full key and update the connection.
  • Model list empty: confirm the connection is enabled and you can reach api.ai24x.com from the host running Open WebUI.
  • VIP named models (e.g. vip-kimi) need an active membership and prepaid credits for those ids.

Why this pairing works

Open WebUI gives your team a shared interface; the gateway decides which model answers it. You get one place to switch tiers, add a VIP named model or control spend — no UI change and no container re-deploy when you do.

Test from the machine that runs Open WebUI

The API connection is made by the Open WebUI process, not by your browser. If the model list stays empty, run the one-line check from the curl guide on the same host — or inside the container itself, if the image ships curl:

docker exec open-webui curl -sS -X POST "https://api.ai24x.com/v1/chat/run" \ -H "X-API-Key: $AI24X_API_KEY" \ -H "Content-Type: application/json" \ -d '{"prompt":"ping","model":"flash","max_tokens":10}'

If that request fails inside the container but works on your laptop, the container has no outbound access to api.ai24x.com — check Docker networking or a corporate proxy before touching the UI settings.

Troubleshooting order

  1. Confirm the connection is enabled and stops at /v1 — nothing appended after it.
  2. Re-paste the full key if you see 401.
  3. Test outbound access from the container as shown above.
  4. Refresh the model list after any connection change.
  5. Only then check model availability for the account, starting with flash.

Create account Cursor guide LobeChat guide Console