in openai create file api, we were using streamed file and the file url from which we were streaming did not had access of streaming, so we were using corrupted file and were giving that fileId to our vector store. and this was the reason for assistants failing to read files.

we resolved this issue by using file buffering instead of streaming.