Connecting AionUi

Difficulty 🧩 Medium　|　API format OpenAI-compatible　|　base_url https://byesu.com/v1

Setup · fill these in

API addresshttps://byesu.com/v1

API keysk-xxxxxxxxCreate in console →

Modelgpt-5.5

API format OpenAI-compatible · model names follow the console

AionUi is a free, local, open-source Cowork desktop app with a graphical interface and a remote WebUI. It can run multiple AI agents at the same time, such as Gemini CLI, Claude Code, and Codex. Point it at byesu — an AI API gateway with OpenAI-compatible and Anthropic-native endpoints — and one key lets you collaborate across models like gpt-5.5, claude-opus-4-8, and gemini-3.1-pro-high in a single interface, pay-as-you-go with no subscription.

Before you start

First go to Console → Tokens and create an API token (it looks like sk-xxxx).

When creating it, be sure to pick the right group — the group determines which models you can call and how you're billed. Not sure which one to choose? See Choosing a group.

Step 1: Download and install AionUi

Go to the GitHub Releases page and download the package for your operating system:

• Download link: https://github.com/iOfficeAI/AionUi/releases
• On Windows choose the .exe, on macOS the .dmg, and on Linux the .AppImage.

After downloading, double-click to install and open AionUi.

Step 2: Add the byesu gateway as a model provider

AionUi connects to OpenAI-compatible services through a unified model configuration, and the byesu gateway falls into this category.

1. Open AionUi and go to the Settings page.
2. Switch to the Model Configuration tab.
3. Click Add Model.
4. For the provider type, choose NewAPI (or OpenAI-compatible / Custom).
5. Fill in the fields as shown below:

Field	What to enter
API Address (base_url)	https://byesu.com/v1
API Key	Your token sk-xxxx
Model	See the recommendations below
Request Protocol	Choose OpenAI (compatible); if a claude / gemini model throws a protocol or format error after being added, manually switch to the protocol that matches the model

The base_url must end with /v1

For the OpenAI-compatible format you must enter https://byesu.com/v1. Don't leave off the /v1, and don't add a trailing slash or /chat/completions at the end.

Once you've entered the key, select the model you want from the model dropdown. If the list is empty, you can type the model name manually. Common models:

Use case	Model name
General chat / writing	gpt-5.5
Coding / long-context reasoning	claude-opus-4-8
Multimodal / very large context	gemini-3.1-pro-high

Save the configuration.

Step 3: (Optional) Configure the Claude Code agent

Each AI agent inside AionUi can be configured independently. The Claude Code agent uses Anthropic's native API, whose base_url differs from the OpenAI-compatible one — it does not include /v1:

Claude Code agent settings

Base URL:  https://byesu.com
API Key:   sk-xxxx
Model:     claude-opus-4-8

Equivalent environment variables

ANTHROPIC_BASE_URL=https://byesu.com
ANTHROPIC_AUTH_TOKEN=sk-xxxx
ANTHROPIC_MODEL=claude-opus-4-8

Other agents like Gemini CLI and Codex work the same way: in each one's settings, enter the matching base_url (use https://byesu.com/v1 for OpenAI-compatible), key, and model.

Step 4: Run it and verify

Go back to the chat interface, select the model you just configured (e.g. gpt-5.5) from the model picker at the top, and send something like "Hi, introduce yourself in one sentence."

If you get a normal reply, the connection is working. If you set up the Claude Code agent, you can ask it to read a file or run a simple task to confirm the agent works.

Stuck?

• 401 / invalid token: Check that the API Key was pasted in full with no extra spaces, and confirm you're using the sk-xxxx created in the byesu console. See Common errors #auth for details.
• "No available channel" / model not found: This usually means the token's group is wrong, or the model name you entered isn't supported by the current group. Switch the group or change the model name. See Choosing a group and Common errors #no-channel.
• Insufficient balance: If your account credit runs out, requests are rejected. Top up or switch groups first. See Common errors #balance.
• Can't connect / 404: Make sure the OpenAI-compatible setting uses https://byesu.com/v1, and the Claude Code agent uses https://byesu.com (without /v1).

For more troubleshooting, see Common errors.