My assistant setup in the OpenAI Playground, linked to a vector store containing text files, functions flawlessly. However, when using the API, the assistant cannot access the documents and fails to reference them in its responses. It responds in proper sentences but does not refer to the files and indicates it cannot access specific files when queried. The vector store and assistant are correctly created and linked, and text files are uploaded and accessible. Headers such as {'OpenAI-Beta': 'assistants=v2'}
are included in the API calls. Despite this, the assistant fails to utilise the files via the API.
Question
Why does the assistant fail to access documents during API calls, despite the correct setup and headers, while it works fine in the Playground?