tomy-kyu

Due to the gRPC migration, it appears that after switching Weaviate to gRPC, the index_node_id returned from the vector database is now being treated as a UUID object rather than a string. This has caused SQLAlchemy to generate UUID-type parameters, and PostgreSQL is failing to perform the type conversion. Consequently, the following error was being output:

Run failed: (psycopg2.errors.UndefinedFunction) operator does not exist: character varying = uuid
LINE 1: ...T_TIMESTAMP WHERE document_segments.index_node_id = 'c6f0cc1...
^
HINT: No operator matches the given name and argument types. You might need to add explicit type casts.

tomy-kyu

I attempted to create core plugin functionality for the Responses API by combining gpt-5.1-codex-max with Gemini3, but indeed, their specifications differ significantly from the Completion API. When attempting to use Vision features, I confirmed that GPT responses become noticeably slower. The responses come back extremely sluggish.

When interactions don't involve images, the system appears to function without issues. Since gpt-5.1-codex is difficult to use in Azure AI Foundry's Chat Playground, having it compatible with Dify would certainly be extremely helpful. In some cases, I might need to narrow down its functionality, but it's also possible that I just lack the necessary technical expertise. I'd like to have someone more experienced review this area to see if it could be improved further.

Using this setup, I've created a plugin for my self-hosted Dify-CE tenant that currently restricts itself to GPT-5.x versions. (I've excluded models like Embedding, TTS, and STT from this implementation.)