REST API

Input formats

The input array is what the user sends in the current turn. Each item is one of three types: text, file or voice.

input is required and contains at least one item. You can mix types in a single array — for example, a text question plus the file it's about.

text

A text prompt — the main format.

{ "type": "text", "text": "Translate the document into English" }

file

A file to analyze. Upload it first via POST /api/files, then put the returned object into input. The model decides on its own which tool to process it with (document reading, image recognition and so on).

{
  "type": "file",
  "file": {
    "id": "42",                         // a string, not a number
    "name": "report.pdf",
    "mime_type": "application/pdf",
    "provider_data": { "path": "…" }
  }
}

voice

A voice message. It's transcribed to text before processing. Upload it the same way as a file.

{
  "type": "voice",
  "voice": {
    "id": "7",
    "name": "message.ogg",
    "mime_type": "audio/ogg",
    "duration": 12.5                     // seconds, optional
  }
}

Example: text + file

"input": [
  { "type": "text", "text": "Summarize the document" },
  { "type": "file", "file": { "id": "42", "name": "report.pdf", "mime_type": "application/pdf" } }
]

id is always a string

In input[].file.id and input[].voice.id the identifier is passed as a string, even if POST /api/files returned it as a number. The id, name and mime_type fields are required.