Teselo Docs

Chat completions

Teselo uses the OpenAI-compatible chat completions format.

Request

Send POST to https://api.beyonditsparts.ai/v1/chat/completions with a bearer API key and Content-Type: application/json.

JSON request

{
  "model": "[MODEL ID — LEARNING]",
  "messages": [
    {
      "role": "user",
      "content": "What is 12 times 8?"
    }
  ]
}

model: The Learning or Frozen model identifier. Exact IDs: [MODEL IDS].
messages: An ordered array of messages. The example uses a user message with text content.

Response

Read the assistant answer from choices[0].message.content. The following is an illustrative OpenAI-compatible shape, not a captured response. Bracketed values are placeholders; timestamps and token counts are numeric in the standard format.

JSON response shape

{
  "id": "[COMPLETION ID]",
  "object": "chat.completion",
  "created": "[UNIX TIMESTAMP]",
  "model": "[MODEL ID]",
  "choices": [
    {
      "index": 0,
      "message": { "role": "assistant", "content": "[ANSWER]" },
      "finish_reason": "[FINISH REASON]"
    }
  ],
  "usage": {
    "prompt_tokens": "[TOKEN COUNT]",
    "completion_tokens": "[TOKEN COUNT]",
    "total_tokens": "[TOKEN COUNT]"
  }
}

Streaming

Streaming: [SUPPORTED? Y/N]

Streaming support has not been confirmed. No streaming example is provided until the host contract is available.

Teselo metadata: x_cl

A response may include an x_cl object. Treat it as optional: an absent object does not identify the answer source.

Teselo response metadata
Field	Meaning
`x_cl.serve_source`	`host-catalog` means the answer came from a proven skill. In the Playground, other values are shown as answered by the model.
Timing fields	[CONFIRM FIELD LIST]

x_cl fields, timing names and units: [CONFIRM FIELD LIST]

No timing field names or full list of source values are assumed here.