Prerequisites

Get a remote model API, set your API key, and verify your endpoint before running QitOS.

Before you run any LLM-powered QitOS example, you need access to a remote model API.

QitOS works well with any OpenAI-compatible provider, as long as you have:

• an account
• an API key
• a base URL
• a model name you can call

Recommended providers

If you are just getting started, these two providers are practical starting points in China and Asia:

• SiliconFlow
• Alibaba Cloud Model Studio / Bailian

Both platforms are commonly used for OpenAI-compatible access, and their signup, pricing, and trial availability can change over time. Check the provider console and pricing pages for the latest terms.

Option 1: SiliconFlow

SiliconFlow’s official quickstart shows this flow:

1. Sign in on the SiliconFlow website.
2. Open the API key page.
3. Create a new API key.
4. Choose a model from the model plaza.
5. Use the OpenAI-compatible base URL:

https://api.siliconflow.cn/v1

Their official quickstart also shows direct use through the OpenAI Python client.

Option 2: Alibaba Cloud Model Studio / Bailian

Alibaba Cloud’s official docs describe this flow:

1. Sign in to Bailian / Model Studio.
2. Open the API Key page in the correct region.
3. Create an API key.
4. Copy the API key and keep it private.
5. Use the OpenAI-compatible base URL for your region.

For the Beijing region, the official compatible endpoint is:

https://dashscope.aliyuncs.com/compatible-mode/v1

Alibaba Cloud’s billing docs also mention that trial offers may be available in the console, such as free trial credit for eligible users. Always verify the latest offer in your account before relying on it.

Set environment variables

Once you have an API key and endpoint, export them before running QitOS:

export OPENAI_API_KEY="sk-..."
export OPENAI_BASE_URL="https://api.siliconflow.cn/v1"

Or for Bailian:

export OPENAI_API_KEY="sk-..."
export OPENAI_BASE_URL="https://dashscope.aliyuncs.com/compatible-mode/v1"

Verify your endpoint

It is worth doing a quick compatibility check before running a larger example:

from openai import OpenAI

client = OpenAI(
    api_key="YOUR_API_KEY",
    base_url="YOUR_BASE_URL",
)

response = client.chat.completions.create(
    model="YOUR_MODEL_NAME",
    messages=[{"role": "user", "content": "Say hello from QitOS."}],
)

print(response.choices[0].message.content)

If this call works, your provider setup is usually ready for QitOS.

What to plug into QitOS

When configuring a QitOS model, you usually need:

• api_key
• base_url
• model

In the simplest OpenAI-compatible case:

from qitos.models import OpenAICompatibleModel

llm = OpenAICompatibleModel(
    model="Qwen/Qwen2.5-72B-Instruct",
    api_key=os.environ["OPENAI_API_KEY"],
    base_url=os.environ["OPENAI_BASE_URL"],
)

Security note

Never commit API keys into the repository, screenshots, or shared trace artifacts. Prefer environment variables, secret managers, or provider-managed temporary credentials.

Next step

Once your endpoint is ready, continue with Quickstart or Build your first agent.