Completions & Responses API

The Completions and Responses APIs provide OpenAI-compatible endpoints for text generation. The /v1/completions endpoint is designed for simple text completion tasks, while the /v1/responses endpoint offers advanced features like asynchronous processing and reasoning support.

Overview

The API provides two main endpoints:

• /v1/completions: OpenAI-compatible text completion endpoint for generating text from a prompt.
• /v1/responses: Advanced responses API with support for asynchronous processing, reasoning, and extended capabilities.

Completions Endpoint

Endpoint

POST https://api.inference.nebul.io/v1/completions

Parameters

Parameter	Type	Required	Description
model	String	Yes	The model ID to use.
prompt	String or Array	Yes	The prompt(s) to complete. Can be a string or array of strings.
temperature	Number	No	Sampling temperature (0.0-2.0). Defaults to 1.0.
top_p	Number	No	Nucleus sampling parameter (0.0-1.0). Defaults to 1.0.
n	Integer	No	Number of completions to generate. Defaults to 1.
stream	Boolean	No	Whether to stream responses. Defaults to false.
stop	String or Array	No	Stop sequences. Can be a string or array of strings.
max_tokens	Integer	No	Maximum number of tokens to generate. Defaults to 16.
presence_penalty	Number	No	Penalty for token presence (-2.0 to 2.0). Defaults to 0.0.
frequency_penalty	Number	No	Penalty for token frequency (-2.0 to 2.0). Defaults to 0.0.
seed	Integer	No	Random seed for reproducible outputs.
echo	Boolean	No	Echo back the prompt in addition to the completion. Defaults to false.
suffix	String	No	Suffix that comes after a completion of inserted text.
best_of	Integer	No	Generates best_of completions and returns the one with the highest log probability.
logprobs	Integer	No	Include log probabilities on the logprobs most likely tokens.
user	String	No	Unique identifier for the end user.

Code Examples

Python

python

1234567891011121314151617

import requests

url = "https://api.inference.nebul.io/v1/completions"
headers = {
    "Authorization": "Bearer <YOUR_API_KEY>",
    "Content-Type": "application/json",
}

payload = {
    "model": "Qwen/Qwen3-30B-A3B-Instruct-2507",
    "prompt": "The capital of France is",
    "max_tokens": 10,
    "temperature": 0.7
}

response = requests.post(url, headers=headers, json=payload)
print(response.json())

cURL

bash

123456789

curl -X POST https://api.inference.nebul.io/v1/completions \
  -H "Authorization: Bearer <YOUR_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "Qwen/Qwen3-30B-A3B-Instruct-2507",
    "prompt": "The capital of France is",
    "max_tokens": 10,
    "temperature": 0.7
  }'

tip

Asynchronous Processing: The /v1/responses endpoint is designed for long-running requests. Use the request ID returned in the response to check status and retrieve results asynchronously, especially for reasoning models that may take longer to process.

tip

Reasoning Models: Some models support extended reasoning capabilities. When using reasoning models, the response may include intermediate reasoning steps and extended metadata beyond standard completions.

Response Format

json

12345678910111213141516171819

{
  "id": "cmpl-abc123",
  "object": "text_completion",
  "created": 1677652288,
  "model": "Qwen/Qwen3-30B-A3B-Instruct-2507",
  "choices": [
    {
      "text": " Paris.",
      "index": 0,
      "logprobs": null,
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 5,
    "completion_tokens": 2,
    "total_tokens": 7
  }
}

Responses Endpoint

The /v1/responses endpoint provides advanced features including asynchronous processing, reasoning support, and extended capabilities beyond the standard completions API.

Endpoint

POST https://api.inference.nebul.io/v1/responses

Parameters

Parameter	Type	Required	Description
model	String	Yes	The model ID to use.
input	String or Array	Yes	The input(s) to process. Can be a string or array of inputs.
temperature	Number	No	Sampling temperature (0.0-2.0). Defaults to 1.0.
top_p	Number	No	Nucleus sampling parameter (0.0-1.0). Defaults to 1.0.
stream	Boolean	No	Whether to stream responses. Defaults to false.
max_output_tokens	Integer	No	Maximum number of tokens to generate. Defaults to 16.
max_tool_calls	Integer	No	Maximum number of tool calls.
user	String	No	Unique identifier for the end user.

Features

The Responses API is designed for more complex use cases that require:

• Asynchronous request processing
• Reasoning model support
• Extended response metadata
• Request cancellation and status tracking

Basic Usage

Python

python

123456789101112131415161718192021222324252627

import requests
import time

url = "https://api.inference.nebul.io/v1/responses"
headers = {
    "Authorization": "Bearer <YOUR_API_KEY>",
    "Content-Type": "application/json",
}

# Create a response request
payload = {
    "model": "Qwen/Qwen3-30B-A3B-Instruct-2507",
    "input": "Explain quantum computing in simple terms.",
    "max_output_tokens": 200
}

response = requests.post(url, headers=headers, json=payload)
response_data = response.json()

# The response includes a response_id for tracking
response_id = response_data.get("id")

# Retrieve the response status
if response_id:
    status_url = f"https://api.inference.nebul.io/v1/responses/{response_id}"
    status_response = requests.get(status_url, headers=headers)
    print(status_response.json())

cURL

bash

12345678910111213

# Create a response request
RESPONSE_ID=$(curl -X POST https://api.inference.nebul.io/v1/responses \
  -H "Authorization: Bearer <YOUR_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "Qwen/Qwen3-30B-A3B-Instruct-2507",
    "input": "Explain quantum computing in simple terms.",
    "max_output_tokens": 200
  }' | jq -r '.id')

# Retrieve the response status
curl -X GET "https://api.inference.nebul.io/v1/responses/${RESPONSE_ID}" \
  -H "Authorization: Bearer <YOUR_API_KEY>"

Retrieving Responses

After creating a response request, you can retrieve the status and results using the response ID:

Endpoint: GET https://api.inference.nebul.io/v1/responses/{response_id}

Python

python

123456789

import requests

url = f"https://api.inference.nebul.io/v1/responses/{response_id}"
headers = {
    "Authorization": "Bearer <YOUR_API_KEY>",
}

response = requests.get(url, headers=headers)
print(response.json())

cURL

bash

12

curl -X GET "https://api.inference.nebul.io/v1/responses/{response_id}" \
  -H "Authorization: Bearer <YOUR_API_KEY>"

Canceling Responses

You can cancel an in-progress response request:

Endpoint: POST https://api.inference.nebul.io/v1/responses/{response_id}/cancel

Python

python

123456789

import requests

url = f"https://api.inference.nebul.io/v1/responses/{response_id}/cancel"
headers = {
    "Authorization": "Bearer <YOUR_API_KEY>",
}

response = requests.post(url, headers=headers)
print(response.json())

cURL

bash

12

curl -X POST "https://api.inference.nebul.io/v1/responses/{response_id}/cancel" \
  -H "Authorization: Bearer <YOUR_API_KEY>"

Model Specifications

The Responses API supports the same LLM models as the Chat Completions API. See the Chat Completions Model Specifications for a complete list of available models.

Reasoning Support

The Responses API supports reasoning models that provide step-by-step reasoning before generating a final answer. See the Reasoning Models guide for detailed information about reasoning-capable models and how to use them.

Best Practices

1. Use Completions for Simple Tasks: Use /v1/completions for straightforward text completion tasks.
2. Use Responses for Complex Workflows: Use /v1/responses when you need asynchronous processing, status tracking, or reasoning support.
3. Polling: When using the Responses API, implement appropriate polling intervals to check response status.
4. Error Handling: Always handle errors and check response status codes.
5. Cancellation: Cancel long-running requests when they're no longer needed to free up resources.