the store parameter in client.chat.completions.create seems to control API call logging, description is not accurate?

[elliscode]

I was trying to get my API calls to show up in the "Logs" pane in the OpenAI API dashboard.

I discovered that setting store=True, does exactly what I want on a per-API call basis, however when reading the description of the parameter in the docstring I see this

openai-python/src/openai/resources/chat/completions/completions.py

Lines 470 to 474 in 41ee03f

	store: Whether or not to store the output of this chat completion request for use in
	our [model distillation](https://platform.openai.com/docs/guides/distillation)
	or [evals](https://platform.openai.com/docs/guides/evals) products.
	Supports text and image inputs. Note: image inputs over 8MB will be dropped.

          store: Whether or not to store the output of this chat completion request for use in
              our [model distillation](https://platform.openai.com/docs/guides/distillation)
              or [evals](https://platform.openai.com/docs/guides/evals) products.

              Supports text and image inputs. Note: image inputs over 8MB will be dropped.

Which seems to be innacurate to me.

Is this flag really controlling whether or not to store the output for model distillation? or is it simply controlling if my request shows up in the "Logs" tab?

Edit: Also wanted to note that this is also the official description in the API reference

https://platform.openai.com/docs/api-reference/chat/create#chat_create-store

[xXMrNidaXx]

The store parameter behavior — good catch on the docs!

What it actually does:

response = client.chat.completions.create(
    model="gpt-4",
    messages=[...],
    store=True  # Stores the request for fine-tuning/analysis
)

When store=True:

• Request/response logged to your OpenAI account
• Available for fine-tuning dataset creation
• Shows in usage dashboard with full content
• Used for evals and analysis

When store=False (default for API):

• Minimal logging
• Content not retained for fine-tuning
• Better for sensitive data

Privacy implications:

# For sensitive data
response = client.chat.completions.create(
    ...,
    store=False  # Don't retain content
)

The docs issue:
Docs say "logging" but it's more about data retention for training/analysis, not just audit logging.

Recommendation:

# Production with PII
store=False

# Building training data
store=True

We manage API compliance at RevolutionAI. This distinction matters for GDPR/HIPAA!

Would be great if docs clarified the retention vs logging difference.

[manimovassagh]

You're right that the docs are somewhat misleading. The store parameter does both things -- it controls whether the request/response pair is retained in your OpenAI account for use in distillation, evals, and whether it shows up in the Logs tab of the dashboard.

They're the same mechanism: when store=True, OpenAI persists the full request and response. The Logs tab, the distillation workflow, and the evals product all read from that same stored data. When store=False (the default for API calls), nothing is retained and nothing shows in Logs.

So the docs are technically accurate -- it stores the data "for use in distillation or evals" -- but they undersell that the most immediately visible effect is that your requests appear in the Logs tab. The Logs tab isn't a separate logging system; it's a view into that same stored data.

Worth noting:

• store defaults to True in the ChatGPT UI but False in the API
• Stored data is retained for 30 days by default (configurable in your org settings)
• If you're working with sensitive data, keep store=False and be aware that even with store=True, OpenAI states stored data is not used to train their models unless you explicitly opt in via fine-tuning