How to connect frontier model APIs: one base_url line

Connecting your workload to Token Spark takes one line of configuration. The API is OpenAI-compatible, so existing SDKs, frameworks and toolchains keep working as is — no need to rewrite business code.

Three steps to connect

Step one: register and create an API key in the console. Step two: point your base_url at the platform endpoint. Step three: set the model name in the request and you are calling.

from openai import OpenAI
client = OpenAI(
    base_url="https://tokenspark.uno/v1",
    api_key="sk-xxxxxxxxxxxx",  # key created in the console
)
resp = client.chat.completions.create(
    model="<model>",  # target model, see the console list
    messages=[{"role": "user", "content": "Hello"}],
)

Once connected, the console shows per-call records: model, token usage, unit price and settled amount — all queryable. It is recommended to run non-critical tasks for a few days first, confirm latency and stability meet expectations, and only then migrate core traffic.

FAQ

Q: How much code do I need to change?

A: Usually one line: point base_url at the platform endpoint and fill in your key, then set the model name as needed.

Q: Is streaming supported?

A: Yes. The API is OpenAI-compatible and the stream parameter behaves the same; third-party clients work too.

Q: What if I hit an error?

A: Check the error-code reference in the docs center first; if it remains unresolved, reach the platform through in-site tickets with the request time and model name.