Seeking Official Support for Custom Embeddings not supported by Fastembed.

Hello qdrant-client Team,

I am reaching out for assistance regarding the integration of a custom model, specifically OpenAI models such as ADA 02, with qdrant-client. Our team has developed a workaround that enables the use of OpenAI models, which are not supported by Fastembed. However, we are keen on adhering to official versions and methods to ensure stability and compatibility.

Since autogen depends on qdrant_client, our current solution falls outside the standard integration process. We are interested in knowing if there is an official or recommended approach to integrate ADA 02 with qdrant-client. Is there a possibility of customizing the retrieval process within the official framework of qdrant-client to accommodate custom embeddings models?

Your guidance on this matter would be invaluable. We would greatly appreciate any suggestions, official documentation, or examples that could assist us in aligning our custom model integration with the official qdrant-client capabilities.

Thank you very much for your time and support.

Best regards,
Andrej

Start by reviewing qdrant-client’s Fastembed integration and the custom retrieval or embedding paths mentioned in the issue. Determine whether OpenAI ADA 02 can be supported through an existing official interface; done would require a decided integration approach, with corresponding documentation or examples if the project accepts the feature.