Chat Completions API - Nbility AI Gateway

POST /v1/chat/completions is the broadest OpenAI-compatible conversation endpoint. Use it for text, multi-turn messages, streaming, and—when supported by the model—vision and tool calls.

Basic request

curl https://api.nbility.ai/v1/chat/completions \
  -H "Authorization: Bearer $NBILITY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.4",
    "messages": [
      {"role": "system", "content": "Keep the answer concise."},
      {"role": "user", "content": "What is an API gateway?"}
    ]
  }'

Use an actual model ID returned by GET /v1/models. Replace the example if it is outside your token's scope.

Common fields

Field	Type	Required	Description
`model`	string	yes	Model ID
`messages`	array	yes	Ordered messages, commonly with `system`, `user`, `assistant`, or `tool` roles
`stream`	boolean	no	When `true`, normally returns incremental SSE events
`temperature`	number	no	Sampling randomness; range and support are model-specific
`top_p`	number	no	Nucleus-sampling parameter
`max_tokens`	integer	no	Maximum output tokens; some models use a corresponding newer field
`tools`	array	no	Tools available to the model
`tool_choice`	string \| object	no	Controls tool selection
`response_format`	object	no	JSON or structured-output configuration, when supported
`user`	string	no	End-user identifier for audit purposes

Nbility preserves compatible request shapes where possible, but providers do not support identical parameters. If a request reports an unsupported parameter, remove it or use the model's native protocol instead of assuming every OpenAI field can be translated.

Python SDK

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["NBILITY_API_KEY"],
    base_url="https://api.nbility.ai/v1",
    timeout=60.0,
    max_retries=2,
)

response = client.chat.completions.create(
    model="gpt-5.4",
    messages=[{"role": "user", "content": "Give me a two-line summary."}],
)
print(response.choices[0].message.content)

Streaming

stream = client.chat.completions.create(
    model="gpt-5.4",
    messages=[{"role": "user", "content": "Tell me a very short story."}],
    stream=True,
)

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

Image input

Use this shape only with a vision-capable model:

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

Production clients should also set connection and overall timeouts plus a bounded retry policy; see Errors and retries.