Skip to content

[Feature]: Request for custom endpoint embedding models. #698

Description

@Autumn0716

What problem are you trying to solve?

Use case / Pain point:
Our application relies on domain-specific embedding models to improve semantic search and retrieval quality. However, the current embedding endpoint only supports predefined models, which limits our ability to use custom-trained or internally hosted embedding models that are optimized for our specific data and terminology.

This creates several challenges: retrieval accuracy is lower for specialized domains, we cannot leverage existing investments in custom embedding models, and we may face compliance or data governance constraints when using third-party hosted models. Supporting custom endpoint embedding models would allow us to integrate our own models while maintaining flexibility, performance, and control over our data pipeline.

Proposed solution

We just need to provider more providers and endpoint for embedding model just like this.

Image

Area

Desktop app

Alternatives considered

Besides,we can provide alternatives endpoints like openai,anthropic,volcengine....

Metadata

Metadata

Assignees

Labels

enhancementImprovement to existing feature

Type

No type

Fields

No fields configured for issues without a type.

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions