weiqingy opened a new issue, #1255: URL: https://github.com/apache/flink-agents/issues/1255
### Search before asking - [x] I searched in the [issues](https://github.com/apache/flink-agents/issues) and found nothing similar. ### Description A blank `model` (`""` or `" "`) on a chat model setup is handled differently depending on the provider and language: | Setup | Blank `model` | |---|---| | Java OpenAI (Completions and Responses), Anthropic, Gemini | Uses the default model | | Java watsonx | Uses the default for `""`, but sends `" "` as is | | Java vLLM | Throws `IllegalArgumentException` | | Java Ollama, Bedrock | Sent as is | | Python vLLM | Raises `ValueError` | | Python OpenAI, Azure OpenAI, Anthropic, Ollama, watsonx, DashScope | Sent as is | When it is sent as is, the error comes from the provider at the first call instead of at setup. A blank model is also treated as not capable of native structured output, so `AUTO` quietly uses the prompt path. Python's `BaseChatModelSetup.model` is a plain field with no validator. `OpenAIChatModelSetup` defaults the model only when it is omitted, not when it is blank. Proposed fix: pick one rule and apply it in the base setup in both languages. Either a blank model means the setup's default, or setup fails with a clear message. Failing at setup seems simpler, since not every provider has a default. ### 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]
