Skip to Main

For developers

One endpoint your tools already speak.

Ember speaks the standard chat-completions API. Use your Ember key with the clients you already have and keep usage, access, and models in one place.

Sign up with a tokenCreate an account to get a keyRead the guide

  1. 01

    Get a key

    Sign in, open Keys, and create a key. Every request sends it as a Bearer token.

  2. 02

    Point your client at Ember

    Set the base URL below in any client that speaks the standard chat-completions API. No other change is needed.

  3. 03

    Send a request

    Pick a model by its Ember name and call chat completions, with streaming if you want it.

Your first request

Base URL

https://api.cabinhub.com.cn/cn/v1

Model names are listed by GET /v1/models with your key. Ember does not publish a model storefront.

These samples run as written with your own key.

curl https://api.cabinhub.com.cn/cn/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer sk-YOUR_KEY" \
  -d '{
    "model": "ember-zenith",
    "messages": [{"role": "user", "content": "Hello!"}],
    "max_tokens": 800
  }'

Streaming

Set stream to true on chat completions to receive standard server-sent events.

Leave room for thinking

Thinking models reason before they answer. Set max_tokens to at least 400, and 2048 is a comfortable default, or the visible answer can come back empty.

Clean responses

Responses from Ember models carry no internal service identifiers, error messages never echo credentials, and every request ID is issued by Ember. If you see anything else, report it as a defect.