External

Trae Setup Guide

This guide is based on Trae 3.3.76 (Windows 10). Other versions may look slightly different.

About third-party compatibility

Any third-party client involves many compatibility differences when integrating with an official API. This guide only demonstrates the basic setup — it cannot guarantee that every model / agent runs flawlessly with zero errors. If a particular model or feature turns out to be incompatible, that's normal; try switching models or adjust per the official docs.

We recommend using each model in its own official product when possible.

Trae supports adding OpenAI-compatible / Claude-compatible custom models. GPT and Claude use different API types, so you add them separately — the API format and request URL differ between them, so mind the distinction.

1. Open Settings

Open Trae and click the Settings (gear ⚙) icon in the top-right corner.

2. Go to Model Management

In the settings page, click the menu (☰) in the top-left, and choose Models from the dropdown.

Under Model Management, click "+ Add Model".

3. Add a GPT / Codex Model

1. Fill in the Custom Config

In the "Add Model" dialog, switch to the Custom Config tab and fill in:

• API Format: choose OpenAI Chat Completions
• Custom Request URL: https://www.cctq.ai/v1 (with /v1, and no trailing slash — Trae automatically appends /chat/completions)
• Model ID: the model you want, e.g. gpt-5.6-sol
• API Key: your token key (sk-***)

Expand Advanced Config and continue:

• Model Series: choose GPT-5 series (Trae optimizes prompts and hyperparameters for the series)
• Display Name: anything you like, e.g. gpt-5.6-sol
• Context Window: input 240000, output 32000 (adjust as needed)
• Tool-call Rounds: 200

Click "Add Model" at the bottom when done.

2. Select the Model and Chat

Once added, you'll see gpt-5.6-sol under Model Management → Custom. Back in the chat panel on the right, click the model picker at the bottom and select it under the Custom Models group.

Send any message (e.g. "hello") — a normal reply means the setup succeeded.

4. Add a Claude Model

Claude uses the Claude API, so you need to add another custom model. The steps are the same as above — only the API format and request URL differ.

1. Fill in the Custom Config

Again click "+ Add Model" → the Custom Config tab, and fill in:

• API Format: choose Anthropic Messages
• Custom Request URL: https://www.cctq.ai (without /v1 here, unlike GPT — Trae automatically appends /v1/messages)
• Model ID: the model you want, e.g. claude-opus-4-8
• API Key: your token key (sk-***)

Click "Add Model" at the bottom when done.

2. Select the Model and Chat

Back in the chat panel on the right, click the model picker at the bottom and select claude-opus-4-8 under the Custom Models group.

Send any message (e.g. "hello") — a normal reply means the setup succeeded.