Chat completions

Send a conversation

curl "$CCSIO_BASE_URL/chat/completions" \
  -H "Authorization: Bearer $CCSIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.8-flash-next",
    "messages": [{"role": "user", "content": "Say hello in one sentence."}]
  }'

The API keeps no conversation between calls: send the whole history in messages every time, oldest first, with the roles system, user, assistant and tool.

Parameters

Parameter	Notes
`model`, `messages`	Required.
`stream`	`true` streams the answer as server-sent events; `false` or absent answers one JSON object.
`temperature`, `top_p`, `seed`	Sampling, passed to the model.
`max_tokens`, `max_completion_tokens`, `stop`	Where the answer ends, passed to the model.
`n`, `presence_penalty`, `frequency_penalty`	Passed to the model.
`response_format`	JSON output, when the model supports it.
`tools`, `tool_choice`	Function calling, see below.
`user`	Your own end-user label, passed to the model.

A parameter the model does not support is refused by the model, and you receive its refusal as a 400.

With "stream": true the answer is a text/event-stream. Each data: line carries one chat.completion.chunk; the stream ends with data: [DONE]. The platform always asks the model for the token counts, so the last chunk before [DONE] has empty choices and a usage object.

curl -N "$CCSIO_BASE_URL/chat/completions" \
  -H "Authorization: Bearer $CCSIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.8-flash-next",
    "stream": true,
    "messages": [{"role": "user", "content": "Write a haiku about the sea."}]
  }'

SSE

data: {"id":"chatcmpl-9b1c4e7f2a6d","object":"chat.completion.chunk","created":1792406148,"model":"qwen3.8-flash-next","choices":[{"index":0,"delta":{"role":"assistant","content":"Waves"},"finish_reason":null}]}

data: {"id":"chatcmpl-9b1c4e7f2a6d","object":"chat.completion.chunk","created":1792406148,"model":"qwen3.8-flash-next","choices":[{"index":0,"delta":{"content":" fold into foam"},"finish_reason":null}]}

data: {"id":"chatcmpl-9b1c4e7f2a6d","object":"chat.completion.chunk","created":1792406148,"model":"qwen3.8-flash-next","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

data: {"id":"chatcmpl-9b1c4e7f2a6d","object":"chat.completion.chunk","created":1792406148,"model":"qwen3.8-flash-next","choices":[],"usage":{"prompt_tokens":16,"completion_tokens":17,"total_tokens":33}}

data: [DONE]

What arrives on the wire.

Tools (function calling)

Describe your functions in tools. When the model decides to call one, the answer carries the calls in choices[0].message.tool_calls; run them and send the results back as tool messages.

import os
from openai import OpenAI

client = OpenAI(
    base_url=os.environ["CCSIO_BASE_URL"],
    api_key=os.environ["CCSIO_API_KEY"],
)

tools = [{
    "type": "function",
    "function": {
        "name": "get_weather",
        "description": "The current weather of a city.",
        "parameters": {
            "type": "object",
            "properties": {"city": {"type": "string"}},
            "required": ["city"],
        },
    },
}]

completion = client.chat.completions.create(
    model="qwen3.8-flash-next",
    messages=[{"role": "user", "content": "What is the weather in Bogotá?"}],
    tools=tools,
)
for call in completion.choices[0].message.tool_calls or []:
    print(call.function.name, call.function.arguments)  # get_weather {"city": "Bogotá"}

Grouping requests

Send X-Session-Id with a label of your own (a conversation, a job, a ticket) to find all its requests together in the request log.