Implementation:Openai Openai python Completion Choice Model

Knowledge Sources	Openai_Openai_python OpenAI API Reference
Domains	API_Types, Python
Last Updated	2026-02-15 00:00 GMT

Overview

Concrete type for a single completion choice within a legacy completion response, provided by the openai-python SDK.

Description

The CompletionChoice Pydantic model represents one generated completion from the legacy completions API. It includes a finish_reason (stop, length, or content_filter), an index indicating its position in the choices list, the generated text, and optional logprobs via the nested Logprobs model. The Logprobs model contains text_offset, token_logprobs, tokens, and top_logprobs arrays for detailed token probability analysis.

Usage

Import this type when you need to type-hint or inspect individual choices from a Completion response object.

Code Reference

Source Location

• Repository: openai-python
• File: src/openai/types/completion_choice.py

Signature

class Logprobs(BaseModel):
    text_offset: Optional[List[int]] = None
    token_logprobs: Optional[List[float]] = None
    tokens: Optional[List[str]] = None
    top_logprobs: Optional[List[Dict[str, float]]] = None

class CompletionChoice(BaseModel):
    finish_reason: Literal["stop", "length", "content_filter"]
    index: int
    logprobs: Optional[Logprobs] = None
    text: str

Import

from openai.types import CompletionChoice

I/O Contract

Fields

Name	Type	Required	Description
finish_reason	Literal["stop", "length", "content_filter"]	Yes	Why the model stopped: natural stop, max tokens, or content filter
index	int	Yes	Position of this choice in the choices list
logprobs	Optional[Logprobs]	No	Token log probability information
text	str	Yes	The generated completion text

Logprobs Fields

Name	Type	Required	Description
text_offset	Optional[List[int]]	No	Character offsets for each token in the text
token_logprobs	Optional[List[float]]	No	Log probabilities for each token
tokens	Optional[List[str]]	No	The individual tokens
top_logprobs	Optional[List[Dict[str, float]]]	No	Top candidate tokens and their log probabilities

Usage Examples

from openai import OpenAI

client = OpenAI()

completion = client.completions.create(
    model="gpt-3.5-turbo-instruct",
    prompt="The capital of France is",
    max_tokens=10,
    logprobs=3,
)

choice = completion.choices[0]
print(choice.text)           # e.g. " Paris"
print(choice.finish_reason)  # e.g. "stop"
if choice.logprobs:
    print(choice.logprobs.tokens)

Related Pages

• Environment:Openai_Openai_python_Python_3_9_Plus