Chat Completions


POST https://api.fenno.ai/v1/chat/completions
Authorization: Bearer <FENNO_API_KEY>
Content-Type: application/json

Request fields

Field	Type	Meaning
`model`	string, required	Exact available Chat Completions model ID
`messages`	array, required	Conversation history; use model-supported roles and content items
`stream`	boolean	Return SSE chunks instead of one completion
`temperature`, `top_p`	number, optional	Sampling controls, only where supported
`max_completion_tokens` or `max_tokens`	integer, optional	Model-dependent output limit; do not assume both are accepted
`tools`, `tool_choice`	array / string or object	Function definitions and selection policy
`response_format`	object, optional	JSON mode or supported JSON Schema configuration
`reasoning_effort`	string, optional	Model-specific reasoning control; not Responses’ `reasoning` object

Python example

Install openai; set FENNO_API_KEY and FENNO_MODEL:


import os
from openai import OpenAI
 
client = OpenAI(
    base_url="https://api.fenno.ai/v1",
    api_key=os.environ["FENNO_API_KEY"],
    timeout=60.0,
    max_retries=0,
)
 
response = client.chat.completions.create(
    model=os.environ["FENNO_MODEL"],
    messages=[
        {"role": "system", "content": "Answer concisely."},
        {"role": "user", "content": "What is an API gateway?"},
    ],
)
print(response.choices[0].message.content)

Send the relevant history on each request. A chat completion ID is not a server-side conversation handle.

Response fields

Field	Meaning
`id`, `model`, `object`	Response identification and model information
`choices[].message`	Assistant text and/or `tool_calls`
`choices[].finish_reason`	Common values include `stop`, `length`, `tool_calls` and safety-related stops
`usage`	Input/output token information when provided

Do not treat every assistant message as plain text: content may be absent during tool calling. An output-limit stop is not a complete structured answer.

Streaming

Set stream=True in the SDK or "stream": true in JSON. Read choices[].delta, not choices[].message. Some chunks contain roles or tool arguments rather than text, and some may have empty choices. The SSE transport ends with [DONE]; inspect finish reasons as well.

stream_options.include_usage is optional and model-dependent. Do not require a usage chunk to render an answer, and do not interpret its absence as zero cost.

See function calling, structured output, streaming and error formats.