Chat Completions

Call OpenGateway’s OpenAI-compatible chat endpoint with a current model ID, inspect request and response fields, and learn where to find model-specific support.

What is the Chat Completions endpoint?#

POST /v1/chat/completions accepts the OpenAI Chat Completions request shape and returns an OpenAI-style completion. OpenGateway routes the request to a provider that can serve the selected model and endpoint.

Model IDs use the creator/model form. Before calling this endpoint, check that the model’s endpoints field from GET /v1/models contains chat_completions. The catalog is public; a restricted API key sees only models allowed for that key.

Endpoint#

POST https://apis.opengateway.ai/v1/chat/completions

Send your API key in Authorization: Bearer <key>.

Request example#

Bash

curl https://apis.opengateway.ai/v1/chat/completions \
  -H "Authorization: Bearer $OPENGATEWAY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek/deepseek-v4.1-flash-ultrafast",
    "messages": [{"role": "user", "content": "Say hello in Korean."}]
  }'

Parameters#

The request follows the Chat Completions shape. Common fields include:

Field	Type	Description
`model`	string	Required catalog model ID in `creator/model` form.
`messages`	array	Required conversation history.
`max_completion_tokens`	integer	Maximum number of generated tokens.
`temperature`, `top_p`, `stop`	number, number, string or array	Sampling controls. Availability and accepted values depend on the model.
`reasoning_effort`	string	Requests a reasoning level where supported. See compatibility for reasoning fields.
`tools`, `tool_choice`	array, string or object	OpenAI-style function tools and selection. See tool calling.
`response_format`	object	Text, JSON object, or JSON Schema output where supported by the model.
`stream`	boolean	Set to `true` to receive server-sent events. See streaming.
`extra.fallbacks`	array of strings	Optional model IDs to try after a retryable failure. See fallbacks.

Supported parameters differ by model. OpenGateway adapts supported fields for the selected provider; consult compatibility for provider-specific behavior.

Response#

A non-streaming response uses the OpenAI Chat Completions response shape. A response contains choices with the assistant message and a finish_reason; token usage is reported in usage.

JSON

{
  "id": "chatcmpl_example",
  "object": "chat.completion",
  "created": 1792728000,
  "model": "deepseek/deepseek-v4.1-flash-ultrafast",
  "choices": [
    {
      "index": 0,
      "message": {"role": "assistant", "content": "안녕하세요!"},
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 14,
    "completion_tokens": 8,
    "total_tokens": 22
  }
}

OpenAI SDK for Python#

Python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["OPENGATEWAY_API_KEY"],
    base_url="https://apis.opengateway.ai/v1",
)

response = client.chat.completions.create(
    model="deepseek/deepseek-v4.1-flash-ultrafast",
    messages=[{"role": "user", "content": "Say hello in Korean."}],
)
print(response.choices[0].message.content)