This will let the users to rely 100% on local OSS models instead of OpenAI Related to: https://github.com/CatchTheTornado/pdf-extract-api/issues/27