weiqingy opened a new issue, #905:
URL: https://github.com/apache/flink-agents/issues/905

   ### Search before asking
   
   - [X] I searched in the 
[issues](https://github.com/apache/flink-agents/issues) and found nothing 
similar.
   
   ### Description
   
   The OpenAI Chat Completions chat model defaults to `gpt-3.5-turbo` when the 
caller omits `model`. This is a legacy OpenAI model — it has been superseded by 
cheaper, more capable small models, and OpenAI itself steers new integrations 
away from it. A user who takes the default today silently gets an outdated 
model.
   
   The stale default lives in both language runtimes and the docs, so a user 
hits it regardless of which API they use:
   
   - Java — `OpenAICompletionsSetup.DEFAULT_MODEL` 
(`integrations/chat-models/openai/src/main/java/org/apache/flink/agents/integrations/chatmodels/openai/OpenAICompletionsSetup.java:61`)
   - Python — `DEFAULT_OPENAI_MODEL` 
(`python/flink_agents/integrations/chat_models/openai/openai_chat_model.py:38`)
   - Docs — the `OpenAICompletionsSetup` parameter tables for both languages 
(`docs/content/docs/development/chat_models.md`)
   
   The Java Javadoc example already uses `gpt-4o-mini` 
(`OpenAICompletionsSetup.java:45`), so the code base is already inconsistent 
with its own default.
   
   Proposed fix: update the default in both runtimes and the docs to 
`gpt-4o-mini` — the current low-cost successor to `gpt-3.5-turbo`, and the 
value the Javadoc example already demonstrates. This keeps a sensible 
zero-config default while dropping the legacy model, and it keeps Java and 
Python in parity.
   
   ### Are you willing to submit a PR?
   
   - [X] I'm willing to submit a PR!
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to