Chatbox

Before you begin

Use the Albatross console to sign in and manage your account. Client API Host uses https://staging-gateway.bulakua.com; it is a different address. Site navigation stays in the current environment. When previewing locally, use the console for the environment you intend to call; keys and usage belong to that same environment. Add /v1 only where the instructions below require it.

Get your Gateway address (the Albatross service address), API Key (your access credential) and exact allowed model name from your service administrator. Log in to the console to check model permissions, available quota and balance. A catalog listing alone does not grant access.

Keep your existing providers and conversations. Add a separate Albatross configuration. A CherryIN, Chatbox AI, ChatGPT or Claude subscription does not fund Albatross requests.

• Open console

1. Download and install

1. Download a stable desktop build for your system from the official site. On macOS open the disk image, install Chatbox, and check Settings → About for the version.
2. Choose your own API provider. Chatbox AI is a separate paid service; this guide does not require its account or subscription.

• Official download
• Official custom-provider guide
• Tested release 1.23.5

2. Configure text chat

1. Open Settings → Model Provider → Add. Name it Albatross and choose OpenAI API Compatible.
2. Set API Host to https://staging-gateway.bulakua.com/v1 and API Path to /chat/completions. Do not enter the console address here. Paste the Albatross key into API Key.
3. Fetch models and add the exact text alias allowed for your key. Check asks you to select a Test Model; this version tests text, vision and tool capabilities in separate requests, so a check can cost more than one request.
4. Create New Chat, select Albatross and <ALBATROSS_MODEL> in the model selector, then send your question. Confirm the reply finishes and the input becomes ready again.

3. Keep and revisit your work

1. Use Copy below the response. The conversation menu → Export Chat → All threads → HTML opens the export dialog; choose your own folder and filename.
2. Keep related follow-ups in the same conversation. Closing and reopening Chatbox retained the tested configuration and conversation. Use Stop to cancel ongoing output, then send a short new request to check recovery.

Work example: a bookstore invitation

1. Start a new chat with <ALBATROSS_MODEL>, the exact text alias allowed for your key. Send: Write a Chinese invitation under 60 characters for the fictional 青禾 bookstore's 春日共读 reading event, Saturday at 3 pm. Remember the event name.
2. Wait for generation to finish, then ask: Translate that invitation into English, keeping the time and event name, under 40 words.
3. Check the date, time and event name yourself. A successful result contains Chinese copy and an English follow-up with the same event details. Copy the approved version into your draft, then export the conversation for your own records.

4. Image Creator: generation only

1. In this version the custom OpenAI-compatible provider is missing from Image Creator. If your built-in OpenAI entry is unused, configure its API Host with the Gateway origin (without /v1) and your Albatross key. Do not overwrite existing OpenAI credentials; use Cherry Studio instead if that entry is occupied.
2. Click Create Image to open Image Creator and explicitly choose OpenAI → GPT Image 2; do not leave a different default model selected. Request one simple square image. The provider label OpenAI describes the client adapter; the address determines that the request goes to Albatross.
3. Click the completed image to open the viewer, then Download to save the PNG. Open the saved file to inspect it.

• Image workflow and limits

Costs and getting help

Albatross Usage and wallet records are the customer bill. Client token counters or price estimates are not your bill; upstream procurement costs are separate. Titles, connection tests and tool continuations may each send paid requests. Stop repeated retries when a request fails.

Invalid Key means checking the Albatross credential. Insufficient quota means checking Albatross balance and limits, not buying a client subscription. For Too Many Requests, wait before retrying; for Gateway Timeout, check Usage before sending again. Both clients displayed these errors in controlled tests; only the invalid-key test reached the local Gateway, with a synthetic key. Quota, rate-limit and timeout responses were simulated without upstream calls.

Contact your Albatross service administrator with the client version, model, time and time zone, error text and request ID if visible. Remove keys, account details and private content from screenshots. Check Usage for keys you can view; ask your organization administrator to check an organization-owned key. Missing records do not mean no charge. Cancelling locally does not guarantee zero upstream work or zero charge.

• Keys, permissions and billing
• Technical error reference

Use one provider with multiple models

1. Keep one Albatross OpenAI API Compatible provider. API Host is https://staging-gateway.bulakua.com/v1, API Path is /chat/completions; enter the key once.
2. Fetch models or manually add exact text aliases from GET /v1/models. In each New Chat choose the model before the first message.
3. If the version cannot change an existing chat's model, create a new chat.

Albatross / <ALBATROSS_TEXT_FAST_ALIAS>
Albatross / <ALBATROSS_TEXT_QUALITY_ALIAS>