Prepare your first API request

Start with a small text request to confirm your application’s authentication and streaming handling. You need a valid key, the confirmed API base URL, an allowed model identifier, and available account allowance.

Request sequence

1. Request GET /api/chat/models with X-API-Key to discover allowed text models.
2. Select a usable model identifier from that result.
3. Optionally send the same prompt and models to POST /api/chat/estimate.
4. Send POST /api/chat/stream with JSON content and accept a server-sent event response.
5. Read complete streamed events until completion or an error.

Example request body:

{
  "prompt": "Explain what a project brief contains in three sentences.",
  "models": ["REPLACE_WITH_AN_ALLOWED_MODEL_ID"],
  "conversation_id": null,
  "enable_web_search": false
}

Replace the placeholder. This example is a request-body reference, not a tested production call.

Field	Meaning
prompt	Nonempty request text
models	Text model IDs; reviewed schema allows one to four
conversation_id	Existing conversation identifier, or null for new work
system_prompt	Optional request-level instructions
enable_web_search	Optional boolean; defaults to false in the reviewed schema
attachment_ids	Optional IDs of already-uploaded attachments; do not assume key-based upload is supported

Check the current public contract before enabling less common output modes. A browser feature and a supported external API operation are not automatically the same thing.