Skip to content

Trae Custom Configuration

This page explains how to connect Trae to this service through a custom OpenAI-compatible configuration.

Trae's exact settings path may vary by version, so this guide focuses on the fields you need rather than one fixed menu path.

1. Open AI / model settings

In Trae settings, look for sections related to:

  • AI
  • Models
  • Provider
  • OpenAI
  • Custom

If your version supports a custom OpenAI-compatible route, it will usually expose fields for a provider, API key, and Base URL.

2. Enter connection details

Use this service's values:

  • Base URL: https://aiapi.aisonnet.org/v1
  • API key: your codex_... key
  • Model ID: use the models returned by GET /v1/models; for a quick connectivity test, try gpt-5.4 first

3. Verify

Save the settings, then start a new chat, edit task, or agent task to confirm Trae can send requests and receive normal responses.

Notes

  • Use the OpenAI-compatible Base URL: https://aiapi.aisonnet.org/v1
  • Do not append the full /chat/completions or /responses endpoint path
  • If your current version only supports built-in official models, or only accepts an official API key without a custom Base URL, it cannot connect to this gateway directly
  • The safest path is to check GET /v1/models first and then enter one of the visible model IDs in Trae

OpenAI-compatible gateway integration docs