Chat Completions

Compatible with OpenAI’s Create chat completion. This is the most general way to call a model: almost every model can be called through it (any model whose endpoint labels include Chat; see Protocol conversion).

POST https://noviahub.com/v1/chat/completions

Headers

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

Request parameters

The table lists the main parameters NoviaHub recognises and forwards. Whether a parameter takes effect depends on the model you call. A model may ignore parameters it doesn’t support, or its upstream may reject them.

Parameter	Type	Required	Description
model	string	Yes	Model ID, such as deepseek-v4-flash. Available IDs are on Models & Pricing.
messages	array	Yes	The conversation; structure below.
stream	boolean	No	true returns a stream (SSE); see Streaming.
stream_options	object	No	Streaming options. include_usage: whether usage is sent at the end; NoviaHub defaults it to true.
max_tokens	integer	No	Maximum tokens to generate.
max_completion_tokens	integer	No	Maximum tokens to generate, including reasoning; OpenAI’s newer name.
temperature	number	No	Sampling temperature; higher is more random.
top_p	number	No	Nucleus sampling.
top_k	integer	No	Top-K sampling (some models).
stop	string / array	No	Stop when any of these strings appears.
n	integer	No	Number of candidate replies.
frequency_penalty	number	No	Frequency penalty.
presence_penalty	number	No	Presence penalty.
seed	number	No	Random seed.
response_format	object	No	Output format, such as {"type": "json_object"} or {"type": "json_schema", "json_schema": {...}}.
tools	array	No	Tools (functions) the model may call, each like {"type": "function", "function": {"name", "description", "parameters", "strict"}}.
tool_choice	string / object	No	Whether and which tool to call.
parallel_tool_calls	boolean	No	Allow several tool calls at once.
reasoning_effort	string	No	Reasoning effort (reasoning models), such as low, medium, high.
logprobs	boolean	No	Return the probability of each token.
top_logprobs	integer	No	How many candidate tokens per position to return probabilities for.
web_search_options	object	No	Web search options (supporting models): search_context_size, user_location.
prompt_cache_key	string	No	Prompt cache key (supporting models).
user, metadata	—	No	User identifier and extra data, forwarded as is.

NoviaHub also recognises and forwards modalities, audio, prediction, logit_bias, store, prompt_cache_retention, verbosity, reasoning, extra_body, and some vendor-specific extensions (such as enable_thinking, thinking_budget, chat_template_kwargs, thinking).

Other fields are dropped

NoviaHub rebuilds the request from this list of fields before sending it upstream, so fields not on the list are not forwarded. service_tier, safety_identifier and stream_options.include_obfuscation are also removed by default.

Structure of messages

Each message is an object:

Field	Description
role	system, developer, user, assistant or tool.
content	A string, or an array mixing text, images and more (below).
name	Optional speaker name.
tool_calls	Tool calls made by the model, in assistant messages.
tool_call_id	In tool messages, which tool call this answers.

When content is an array, each item’s type can be text (with text), image_url (with image_url: {"url": "...", "detail": "..."}), input_audio, file or video_url. Whether a model can read images or audio is shown under Input types on its Models & Pricing page.

Sample code

cURL

curl https://noviahub.com/v1/chat/completions \
  -H "Authorization: Bearer $NOVIAHUB_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [
      {"role": "system", "content": "You are a concise assistant."},
      {"role": "user", "content": "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"],
)

completion = client.chat.completions.create(
    model="deepseek-v4-flash",
    messages=[
        {"role": "system", "content": "You are a concise assistant."},
        {"role": "user", "content": "Explain what an API is in one sentence."},
    ],
)

print(completion.choices[0].message.content)

TypeScript

import OpenAI from 'openai'

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

const completion = await client.chat.completions.create({
  model: 'deepseek-v4-flash',
  messages: [
    { role: 'system', content: 'You are a concise assistant.' },
    { role: 'user', content: 'Explain what an API is in one sentence.' },
  ],
})

console.log(completion.choices[0].message.content)

Response

Field	Description
id	ID of this reply.
object	Always chat.completion.
created	Creation time (Unix seconds).
model	Model ID.
choices[].message.content	The model’s answer.
choices[].message.tool_calls	Tool calls requested by the model, if any.
choices[].message.reasoning_content	Some models return their reasoning here.
choices[].finish_reason	Why generation stopped, such as stop (finished), length (hit the limit), tool_calls (wants to call a tool).
usage.prompt_tokens	Input tokens.
usage.completion_tokens	Output tokens.
usage.total_tokens	Total.

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

{
  "id": "chatcmpl-mock123",
  "object": "chat.completion",
  "created": 1790609866,
  "model": "deepseek-v4-flash",
  "choices": [
    {
      "index": 0,
      "message": { "role": "assistant", "content": "Hello from the mock upstream." },
      "finish_reason": "stop"
    }
  ],
  "usage": { "prompt_tokens": 12, "completion_tokens": 7, "total_tokens": 19 }
}

When a model is reached through protocol conversion (for example a Claude or Gemini model called here), usage contains extra internal fields you can ignore; see Protocol conversion.

Streaming

Add "stream": true and the response arrives as SSE (Server-Sent Events). Each piece starts with data: and holds a chat.completion.chunk object; new text is in choices[0].delta.content. The last line is data: [DONE].

By default NoviaHub sends one more piece before [DONE] that only carries usage (an empty choices array plus usage), even if you didn’t send stream_options. To leave it out, send "stream_options": {"include_usage": false}.

Example stream (test instance)

data: {"id":"chatcmpl-mock123","object":"chat.completion.chunk","created":1790609866,"model":"deepseek-v4-flash","choices":[{"index":0,"delta":{"content":"Hello","role":"assistant"},"finish_reason":null}]}

data: {"id":"chatcmpl-mock123","object":"chat.completion.chunk","created":1790609866,"model":"deepseek-v4-flash","choices":[{"index":0,"delta":{"content":" from the mock upstream."},"finish_reason":null}]}

data: {"id":"chatcmpl-mock123","object":"chat.completion.chunk","created":1790609866,"model":"deepseek-v4-flash","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

data: {"id":"chatcmpl-mock123","object":"chat.completion.chunk","created":1790609866,"model":"deepseek-v4-flash","choices":[],"usage":{"prompt_tokens":12,"completion_tokens":7,"total_tokens":19}}

data: [DONE]

cURL

curl https://noviahub.com/v1/chat/completions \
  -H "Authorization: Bearer $NOVIAHUB_API_KEY" \
  -H "Content-Type: application/json" \
  -N \
  -d '{
    "model": "deepseek-v4-flash",
    "stream": true,
    "messages": [{"role": "user", "content": "Write a four-line poem."}]
  }'

Python

import os

from openai import OpenAI

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

stream = client.chat.completions.create(
    model="deepseek-v4-flash",
    messages=[{"role": "user", "content": "Write a four-line poem."}],
    stream=True,
)

for chunk in stream:
    if chunk.choices and chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="", flush=True)
    if chunk.usage:
        print("\nUsage:", chunk.usage)

TypeScript

import OpenAI from 'openai'

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

const stream = await client.chat.completions.create({
  model: 'deepseek-v4-flash',
  messages: [{ role: 'user', content: 'Write a four-line poem.' }],
  stream: true,
})

for await (const chunk of stream) {
  const text = chunk.choices[0]?.delta?.content
  if (text) process.stdout.write(text)
  if (chunk.usage) console.log('\nUsage:', chunk.usage)
}

The usage-only piece has an empty choices array, so check it before reading choices[0], as the samples do.

Common errors

HTTP status	Cause
400	Missing model (Model name not specified...), missing messages (field messages is required), or the body isn’t valid JSON.
401	The key is invalid, disabled, expired or out of quota (Invalid token).
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 (No available channel for model ...).

See Errors and troubleshooting for details.