curl --request POST \
--url https://api.omniall.ai/v1/responses \
--header 'Authorization: Bearer <token>' \
--header 'Content-Type: application/json' \
--data '
{
"model": "gpt-5",
"input": "Explain Kubernetes in one paragraph"
}
'
{
"id": "<string>",
"object": "response",
"created_at": 123,
"status": "completed",
"model": "<string>",
"output": [
{
"type": "<string>",
"id": "<string>",
"status": "<string>",
"role": "<string>",
"content": [
{
"type": "<string>",
"text": "<string>"
}
]
}
],
"usage": {
"input_tokens": 123,
"output_tokens": 123,
"total_tokens": 123
}
}{
"error": {
"message": "<string>",
"type": "new_api_error",
"code": "<string>"
}
}{
"error": {
"message": "无效的令牌 (request id: ...)",
"type": "new_api_error",
"code": ""
}
}{
"error": {
"message": "<string>",
"type": "new_api_error",
"code": "<string>"
}
}{
"error": {
"message": "<string>",
"type": "new_api_error",
"code": "<string>"
}
}{
"error": {
"message": "<string>",
"type": "new_api_error",
"code": "<string>"
}
}Responses
Responses API
OpenAI Responses API relayed by Omniall AI.
Base path: POST https://api.omniall.ai/v1/responses
Prefer this format for newer OpenAI models, built-in tools (web search), and multi-turn previous_response_id flows.
Scenarios
- Vision / files:
inputitemsinput_image/input_file - Web search: built-in tool
{ "type": "web_search_preview" } - Agent: function tools in
tools - Reasoning:
reasoning.effort
Official reference: https://platform.openai.com/docs/api-reference/responses
POST
/
v1
/
responses
curl --request POST \
--url https://api.omniall.ai/v1/responses \
--header 'Authorization: Bearer <token>' \
--header 'Content-Type: application/json' \
--data '
{
"model": "gpt-5",
"input": "Explain Kubernetes in one paragraph"
}
'
{
"id": "<string>",
"object": "response",
"created_at": 123,
"status": "completed",
"model": "<string>",
"output": [
{
"type": "<string>",
"id": "<string>",
"status": "<string>",
"role": "<string>",
"content": [
{
"type": "<string>",
"text": "<string>"
}
]
}
],
"usage": {
"input_tokens": 123,
"output_tokens": 123,
"total_tokens": 123
}
}{
"error": {
"message": "<string>",
"type": "new_api_error",
"code": "<string>"
}
}{
"error": {
"message": "无效的令牌 (request id: ...)",
"type": "new_api_error",
"code": ""
}
}{
"error": {
"message": "<string>",
"type": "new_api_error",
"code": "<string>"
}
}{
"error": {
"message": "<string>",
"type": "new_api_error",
"code": "<string>"
}
}{
"error": {
"message": "<string>",
"type": "new_api_error",
"code": "<string>"
}
}Authorizations
Use Authorization: Bearer sk-... from https://omniall.ai/dashboard
Body
application/json
OpenAI Responses API request (POST /v1/responses). Prefer this for newer OpenAI models and built-in tools.
Model id from Model Square.
Example:
"gpt-5"
User input: a plain string, or an array of items/parts (input_text, input_image, input_file, etc.).
System-level instructions for this response (similar to a system prompt).
Maximum tokens the model may generate.
Sampling temperature.
Nucleus sampling parameter.
If true, stream response events over SSE.
Streaming options passthrough.
Built-in tools (e.g. { "type": "web_search_preview" }) and/or function tools for agents.
How tools are selected (auto, required, or specific tool).
Allow multiple tool calls in one turn when supported.
Cap on tool calls for this response.
Continue from a prior response id (multi-turn without resending full history).
Conversation object/id when using Responses conversation state.
Reasoning controls for reasoning models.
Show child attributes
Show child attributes
Text format configuration (including structured output / json_schema when supported).
Extra fields to include in the response payload.
How to handle context overflow: auto or disabled.
Available options:
auto, disabled Arbitrary metadata attached to the response.
Whether upstream may store the interaction (channel setting dependent).
End-user identifier for abuse monitoring.
Response
Response object (or SSE events when streaming).
Responses API result object.
Response id (use with previous_response_id).
Usually response.
Example:
"response"
Unix timestamp.
Lifecycle status: completed, failed, in_progress, incomplete.
Available options:
completed, failed, in_progress, incomplete Model used.
Output items (messages, tool calls, etc.). Read message content parts for the answer text.
Show child attributes
Show child attributes
Token usage.
Show child attributes
Show child attributes