Support for OpenAI Prompt Management API (prompt_id) in LangChain

[torrespro]

Checked

• I searched existing ideas and did not find a similar one
• I added a very descriptive title
• I've clearly described the feature request and motivation for it

Feature request

Hi LangChain team,

I wanted to ask if there is any plan or ongoing work to support the new OpenAI Prompt Management API https://community.openai.com/t/enhanced-prompt-management/1290305 in LangChain, specifically the ability to call OpenAI chat completions using saved prompt templates via the prompt_id parameter.

The OpenAI documentation here:
https://platform.openai.com/docs/guides/text?api-mode=responses#reusable-prompts

describes a new feature that allows versioning and managing prompts directly in OpenAI's platform and referencing them via prompt_id in the API call.

Currently, it seems LangChain's OpenAI or ChatOpenAI classes do not support passing a prompt_id to the API, and I wanted to know:

• Are there plans to add native support for this feature soon?
• Any recommended workarounds in the meantime to integrate OpenAI prompt management with LangChain?

Thanks for the great work on LangChain! Looking forward to your feedback.

Best regards

Motivation

Using the Prompt Management API would allow better prompt version control, centralized management, and easier experimentation with prompt iterations without changing client code. This could help LangChain users maintain high-quality prompts and improve collaboration in teams.

Proposal (If applicable)

A possible approach could be to extend the existing OpenAI or ChatOpenAI class to accept a prompt_id parameter, and modify the API call to include it instead of passing raw prompt strings. This might be implemented as a new parameter with backward compatibility, allowing users to opt-in to prompt management features.