SUPREME Setu API — Early Access
The SUPREME Setu OpenAI-compatible endpoints preview the upcoming public release. Want early access or custom organizational quota limits? Contact our solutions engineering team.
Streaming & Reasoning Trace
Setu delivers streaming tokens via standard HTTP Server-Sent Events (SSE). When using reasoning models like quantum-reasoner, thought process traces are streamed prior to the final conversational answer.
Protocol Structure
When "stream": true is passed to POST /chat/completions:
- The HTTP response headers include:
http
Content-Type: text/event-stream; charset=utf-8 Cache-Control: no-cache, no-transform Connection: keep-alive - Each event begins with
data:followed by a JSON chunk. - The final frame is always:
text
data: [DONE]
Chunk Format
json
{
"id": "chatcmpl-setu-1",
"object": "chat.completion.chunk",
"created": 1728259200,
"model": "quantum-lite",
"choices": [
{
"index": 0,
"delta": {
"content": "Hello"
},
"finish_reason": null
}
]
}Reasoning Trace Chunking
For quantum-reasoner, initial chunks contain reasoning logic inside delta.reasoning_content:
json
{
"choices": [
{
"index": 0,
"delta": {
"reasoning_content": "To solve 3x + 12 = 45, subtract 12 from both sides..."
}
}
]
}Once reasoning concludes, standard delta.content chunks stream the final response.
