API reference: Chat Completions

POST /v1/chat/completions answers a list of messages. It follows the OpenAI Chat Completions API. For examples, read the Chat Completions guide. New projects can also use Responses; both share keys, tiers and limits.

POST https://dummydomain/v1/chat/completions
Authorization: Bearer YOUR_API_KEY
Content-Type: application/json

Request body

Field	Type	Description
model	string, required	A tier: Instant, Fast, Smart or Genius
messages	array, required	Messages with roles system, developer, user, assistant and tool. content is a string or a list of text and image_url parts
max_tokens, max_completion_tokens	integer	The longest answer you want
temperature, top_p	number	Sampling. Use one of them
presence_penalty, frequency_penalty	number	Discourage repeats
stop	string or array	Text that ends the answer
seed	integer	Repeatable sampling where possible
stream	boolean	true sends chunks as the answer is written
response_format	object	text, json_object or json_schema. See Structured JSON output
tools, tool_choice, parallel_tool_calls		Function tools on tiers whose capabilities.tools is true

Images go in image_url parts as data: URLs: up to 8 per request, 10 MiB each and 20 MiB in total. An https image address works only if your administrator has allowed its host; this is off by default, and the server then fetches the image itself, with no redirects.

{
  "model": "Fast",
  "messages": [
    {"role": "system", "content": "Reply in one sentence."},
    {"role": "user", "content": "What is GST?"}
  ]
}

Function calls

When the model wants a function, the assistant message has tool_calls. Run each one and send its result as a tool message with the same tool_call_id:

{"role": "tool", "tool_call_id": "call_8f2c...", "content": "{\"status\": \"shipped\"}"}

The completion object

{
  "id": "chatcmpl-57b769...",
  "object": "chat.completion",
  "created": 1791959603,
  "model": "Fast",
  "choices": [
    {
      "index": 0,
      "message": {"role": "assistant", "content": "GST is India's tax on the supply of goods and services."},
      "finish_reason": "stop"
    }
  ],
  "usage": {"prompt_tokens": 24, "completion_tokens": 13, "total_tokens": 37}
}

finish_reason is stop, length (the answer reached the token limit) or tool_calls.

Stream chunks

With "stream": true you get chat.completion.chunk objects. Each delta adds to the answer, and the stream ends with data: [DONE].

data: {"id":"chatcmpl-57b769...","object":"chat.completion.chunk","created":1791959603,"model":"Fast","choices":[{"index":0,"delta":{"content":"GST is"},"finish_reason":null}]}

data: [DONE]

Related

• Chat Completions guide
• Models and Errors

Common questions

Is ZenithAI's /v1/chat/completions the same as OpenAI's?

It follows the same request and response format, so code written for OpenAI chat completions works after you change the base URL and use a tier name such as Fast as the model.