Skip to main content

API Quickstart

info

The page and service rules are subject to the current Console.

First, confirm these three things​

  1. Create an API key on the API Console "Tokens" page, and note which group the key belongs to.
  2. Get the current full model IDs from the Model Marketplace or /v1/models.
  3. The Base URL is fixed at https://llmapi.cttai.art/v1 (OpenAI-compatible).

The examples below use the local environment variable CTTAI_API_KEY — never commit a real key into your repository.

Query models visible to your key​

curl https://llmapi.cttai.art/v1/models \
  -H "Authorization: Bearer $CTTAI_API_KEY"

The model list shows what is currently visible; the exact protocol, modality, and real-time availability are subject to the model description.

Minimal Chat Completions request​

Replace YOUR_AVAILABLE_MODEL with a full model ID that supports this protocol.

curl https://llmapi.cttai.art/v1/chat/completions \
  -H "Authorization: Bearer $CTTAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "YOUR_AVAILABLE_MODEL",
    "messages": [{"role": "user", "content": "Please just reply OK"}]
  }'

This request may incur a small cost. After success, confirm the model and usage in the Console "Usage Logs".

SDK examples​

Node.js (OpenAI SDK)​

npm install openai
import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: process.env.CTTAI_API_KEY,
  baseURL: 'https://llmapi.cttai.art/v1',
});

const result = await client.chat.completions.create({
  model: process.env.CTTAI_MODEL,
  messages: [{role: 'user', content: 'Please just reply OK'}],
});
console.log(result.choices[0].message.content);

Python (OpenAI SDK)​

python -m pip install openai
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ['CTTAI_API_KEY'],
    base_url='https://llmapi.cttai.art/v1',
)
result = client.chat.completions.create(
    model=os.environ['CTTAI_MODEL'],
    messages=[{'role': 'user', 'content': 'Please just reply OK'}],
)
print(result.choices[0].message.content)

Streaming​

Add "stream": true to the request body to receive an SSE stream:

data: {"id":"chatcmpl-...","choices":[{"delta":{"content":"Hello"},"index":0}]}
data: [DONE]

Pre-launch checklist​

  • Use a dedicated key with its own budget and confirm group permissions.
  • Handle connection, read, and business timeouts; for long tasks, check status first and avoid duplicate submissions.
  • Use bounded retry with backoff for transient 429 / 5xx; never blindly retry auth or parameter errors.
  • Log time, model, request id / task id, and the error body; never store a full key in logs.
  • Actual costs are subject to the Console usage logs and billing rules.

Other models​