Chat Completions

Use Chat Completions for text chat, vision input, streaming, JSON output, and tool calling.

POST https://api.rout.my/v1/chat/completions

Request

curl https://api.rout.my/v1/chat/completions \
  -H "Authorization: Bearer $ROUTMY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "openai/gpt-5.4",
    "messages": [
      { "role": "system", "content": "Answer briefly." },
      { "role": "user", "content": "What does this endpoint do?" }
    ],
    "temperature": 0.7,
    "max_tokens": 256
  }'

Request fields

Field	Type	Required	Notes
model	string	Yes	Exact model ID from /v1/models.
messages	array	Yes	Conversation messages with role and content.
stream	boolean	No	Set true for Server-Sent Events.
temperature	number	No	Sampling temperature.
top_p	number	No	Nucleus sampling.
max_tokens	integer	No	Maximum generated tokens.
max_completion_tokens	integer	No	OpenAI-compatible alias for maximum generated tokens.
max_output_tokens	integer	No	Accepted for Gemini-style clients.
stop	string or array	No	Stop sequence or sequences.
logprobs / top_logprobs	boolean / integer	No	Forwarded when the upstream supports log probability output.
store / service_tier	boolean / string	No	Forwarded for OpenAI-compatible upstreams.
prompt_cache_key / prompt_cache_retention	string	No	Forwarded for OpenAI-compatible prompt caching.
safety_identifier	string	No	Accepted for client compatibility but stripped before upstream requests.
modalities / audio	array / object	No	Forwarded for multimodal output-capable upstreams.
image_config	object	No	Image aspect ratio and resolution for image-capable chat models; see below.
prediction	object	No	Forwarded for providers that support predicted outputs.
response_format	object	No	OpenAI-style response format, for example JSON object mode.
tools	array	No	OpenAI-style function tools.
tool_choice	string or object	No	Controls tool selection.
parallel_tool_calls	boolean	No	Controls parallel tool calls when supported.
reasoning	object	No	Provider-compatible reasoning controls when supported.
reasoning_effort	string	No	OpenAI-style reasoning effort; mapped to internal reasoning controls.
metadata	object	No	Client metadata passed through when supported.

Unknown fields are preserved and forwarded when the upstream provider accepts them.

Message content

Text-only messages can use a string:

{
  "role": "user",
  "content": "Summarize this in one paragraph."
}

Vision-capable models can receive an array of content parts:

{
  "role": "user",
  "content": [
    { "type": "text", "text": "Describe this image." },
    {
      "type": "image_url",
      "image_url": {
        "url": "https://example.com/image.png"
      }
    }
  ]
}

Data URLs are accepted for image inputs:

data:image/png;base64,...

Streaming

Set stream to true to receive Server-Sent Events:

{
  "model": "openai/gpt-5.4",
  "messages": [{ "role": "user", "content": "Write three short lines." }],
  "stream": true
}

The stream ends with:

data: [DONE]

Tool calling

{
  "model": "openai/gpt-5.4",
  "messages": [
    { "role": "user", "content": "What is the status of order 123?" }
  ],
  "tools": [
    {
      "type": "function",
      "function": {
        "name": "get_order",
        "description": "Look up an order by ID.",
        "parameters": {
          "type": "object",
          "properties": {
            "order_id": { "type": "string" }
          },
          "required": ["order_id"]
        }
      }
    }
  ],
  "tool_choice": "auto"
}

Image-capable chat models

Some models generate images through the chat endpoint. For those models, include image output in modalities:

{
  "model": "provider/image-chat-model-id",
  "messages": [
    { "role": "user", "content": "Create a square icon of a blue glass cube." }
  ],
  "modalities": ["image", "text"],
  "image_config": {
    "aspect_ratio": "1:1",
    "image_size": "1K"
  }
}

Generated images can appear on the assistant message in an images array with image URLs or data URLs.

image_config

Pass this object at the top level of the /v1/chat/completions request, alongside model and messages.

Field	Type	Notes
aspect_ratio	string	Common values include 1:1, 16:9, 9:16, 4:3, 3:4, 4:5, 5:4, and 21:9.
image_size	string	Common values include 1K, 2K, and 4K. Values are forwarded to the provider when supported.

Both fields are optional. For example, "image_config": {"aspect_ratio": "16:9", "image_size": "2K"} requests a landscape image at 2K resolution. Resolution is separate from the provider's quality hint.

For image models routed through the Images API, all message text is joined into one prompt. Image references from the conversation select image editing; the same image_config is forwarded for generation and editing. With stream: true, completed images arrive in delta.images after generation finishes. Images returned through this bridge use image_url.url: "data:image/<type>;base64,..." in both JSON and SSE; temporary provider storage links are downloaded by GoRouter.

Response

{
  "id": "chatcmpl_abc123",
  "object": "chat.completion",
  "created": 1744000000,
  "model": "openai/gpt-5.4",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "This endpoint creates model responses from chat messages."
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 28,
    "completion_tokens": 12,
    "total_tokens": 40
  }
}