Responses

Compatible with OpenAI’s Create a model response. Tools such as Codex use this endpoint.

POST https://noviahub.com/v1/responses

Only for models labelled Response

Use this endpoint only with models whose endpoint labels include Response (check on Models & Pricing). For models labelled only Chat, NoviaHub does not convert Responses requests to the Chat format; whether they succeed depends on the upstream, and we don’t guarantee it. See Protocol conversion.

Headers

Header	Required	Description
Authorization	Yes	Bearer <your API key>
Content-Type	Yes	application/json

Request parameters

These are the parameters NoviaHub recognises and forwards. Whether a parameter takes effect depends on the model.

Parameter	Type	Required	Description
model	string	Yes	Model ID, such as gpt-6-sol.
input	string / array	No	Input: a string, or an array of messages or content items as defined by OpenAI.
instructions	string	No	System-level instructions.
max_output_tokens	integer	No	Maximum tokens to generate.
temperature	number	No	Sampling temperature.
top_p	number	No	Nucleus sampling.
stream	boolean	No	true returns a stream (SSE).
stream_options	object	No	Streaming options.
tools	array	No	Available tools.
tool_choice	string / object	No	How tools are chosen.
parallel_tool_calls	boolean	No	Allow several tool calls at once.
max_tool_calls	integer	No	Maximum number of tool calls.
reasoning	object	No	Reasoning settings, such as {"effort": "medium", "summary": "auto"}.
text	object	No	Text output settings (format and so on).
previous_response_id	string	No	Continue from an earlier reply.
include	array	No	Extra content to return.
store	boolean	No	Whether the upstream stores this reply.
truncation	string	No	What to do when the context is too long.
prompt_cache_key	string	No	Prompt cache key.
top_logprobs	integer	No	Number of candidate token probabilities to return.
metadata, user	—	No	Extra data and user identifier, forwarded as is.

NoviaHub also recognises conversation, context_management, prompt, prompt_cache_options, prompt_cache_retention, frequency_penalty, presence_penalty, moderation, client_metadata, and extension fields such as enable_thinking, thinking_budget, chat_template_kwargs, top_k, min_p, repetition_penalty and stop.

Fields that are not forwarded

• Fields not on the list are dropped. In our tests, background was not forwarded to the upstream.
• service_tier and safety_identifier are removed by default.

Sample code

cURL

curl https://noviahub.com/v1/responses \
  -H "Authorization: Bearer $NOVIAHUB_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-6-sol",
    "input": "Explain what an API is in one sentence."
  }'

Python

import os

from openai import OpenAI

client = OpenAI(
    base_url="https://noviahub.com/v1",
    api_key=os.environ["NOVIAHUB_API_KEY"],
)

response = client.responses.create(
    model="gpt-6-sol",
    input="Explain what an API is in one sentence.",
)

print(response.output_text)

TypeScript

import OpenAI from 'openai'

const client = new OpenAI({
  baseURL: 'https://noviahub.com/v1',
  apiKey: process.env.NOVIAHUB_API_KEY,
})

const response = await client.responses.create({
  model: 'gpt-6-sol',
  input: 'Explain what an API is in one sentence.',
})

console.log(response.output_text)

Response

Field	Description
id	ID of this reply.
object	Always response.
created_at	Creation time (Unix seconds).
status	Status; completed when finished normally.
model	Model ID.
output	Output items. The answer is in items of type message, in content[] entries of type output_text, under text.
usage.input_tokens / output_tokens / total_tokens	Input, output and total tokens.
usage.input_tokens_details.cached_tokens	Input tokens served from cache.
usage.output_tokens_details.reasoning_tokens	Tokens spent on reasoning.

Example response (test instance; content from a simulated upstream)

{
  "id": "resp_mock123",
  "object": "response",
  "created_at": 1790609867,
  "status": "completed",
  "model": "gpt-6-sol",
  "output": [
    {
      "type": "message",
      "id": "msg_mock123",
      "status": "completed",
      "role": "assistant",
      "content": [{ "type": "output_text", "text": "Hello from the mock upstream.", "annotations": [] }]
    }
  ],
  "usage": {
    "input_tokens": 12,
    "output_tokens": 7,
    "total_tokens": 19,
    "input_tokens_details": { "cached_tokens": 0 },
    "output_tokens_details": { "reasoning_tokens": 0 }
  }
}

Note

NoviaHub does not support fetching an earlier reply with GET /v1/responses/{id}. In our tests it returned 404 No response found with id ....

Streaming

With "stream": true, the response arrives as SSE events. Each event has an event: line and a data: line. Common events:

• response.created: generation started.
• response.output_text.delta: a new piece of text, in delta.
• response.completed: generation finished; response holds the full result, including usage.

Example stream (test instance)

event: response.created
data: {"type":"response.created","sequence_number":0,"response":{"id":"resp_mock123","object":"response","created_at":1790609867,"status":"in_progress","model":"gpt-6-sol","output":[],"usage":null}}

event: response.output_text.delta
data: {"type":"response.output_text.delta","sequence_number":1,"item_id":"msg_mock123","output_index":0,"content_index":0,"delta":"Hello"}

event: response.output_text.delta
data: {"type":"response.output_text.delta","sequence_number":2,"item_id":"msg_mock123","output_index":0,"content_index":0,"delta":" from the mock upstream."}

event: response.completed
data: {"type":"response.completed","sequence_number":3,"response":{"id":"resp_mock123","object":"response","created_at":1790609867,"status":"completed","model":"gpt-6-sol","output":[{"type":"message","id":"msg_mock123","status":"completed","role":"assistant","content":[{"type":"output_text","text":"Hello from the mock upstream.","annotations":[]}]}],"usage":{"input_tokens":12,"output_tokens":7,"total_tokens":19,"input_tokens_details":{"cached_tokens":0},"output_tokens_details":{"reasoning_tokens":0}}}}

Responses Compact

POST https://noviahub.com/v1/responses/compact

This matches OpenAI’s conversation compaction endpoint: it condenses a long conversation history so the conversation can continue. It is available only for models labelled openai-response-compact.

• model is required. Only these fields are forwarded upstream: model, input, instructions, previous_response_id, parallel_tool_calls, prompt_cache_key, prompt_cache_options, prompt_cache_retention.
• tools, reasoning and text are recognised but not forwarded; service_tier is removed by default.
• No streaming; the response always comes back in one piece.
• The body is returned from the upstream as is, with id, object, created_at, output and usage (plus error on failure).
• Billed by tokens, like a normal chat request.

/v1/alpha/search

POST https://noviahub.com/v1/alpha/search

This is the standalone web search endpoint used by the Codex command-line tool; most applications don’t need it. It is available only for models labelled openai-alpha-search.

• Only model is required. The body is forwarded upstream as is, and the response is returned as is.
• No streaming.
• Billing is different: the upstream reports no token usage, so NoviaHub charges one web-search tool call per request, with zero tokens. The price is set by the platform.
• Calling it for an unsupported model returns an error, such as channel does not support /v1/alpha/search.

Common errors

HTTP status	Cause
400	Missing model, or the body isn’t valid JSON.
401	The key is invalid, disabled, expired or out of quota.
403	The key isn’t allowed to use this model, the IP isn’t on the allow list, or the account balance is too low.
503	Wrong model ID, or no channel is currently available for the model.

See Errors and troubleshooting for details.