Feature Request: Add Support for Embedding Models

[celsofranssa]

Summary

I'd like to request support for embedding models on the OpenRouter platform.

Motivation

Adding embedding models would significantly enhance the utility of OpenRouter for a wide range of use cases beyond simple text generation. Specifically, it would enable developers to build and deploy applications that rely on:

• Retrieval-Augmented Generation (RAG): Embedding models are a foundational component of RAG, allowing for semantic search over private or public datasets to provide context to a language model.
• Semantic Search: Building search engines that understand the meaning and intent of a query, rather than just keywords.
• Clustering and Classification: Grouping and organizing data based on semantic similarity.
• Recommendation Systems: Finding and recommending content that is semantically similar to a user's interests.

Suggested Models

While any embedding model would be a great start, I would suggest considering popular and high-performance options such as:

• OpenAI's text-embedding-3-small and text-embedding-3-large
• Cohere's embed-english-v3.0
• Open-source models like BAAI/bge-large-en or models from the huggingface/sentence-transformers library.

Why OpenRouter?

Integrating these models would make OpenRouter an even more comprehensive and powerful tool for developers, as they could access both language and embedding models through a single, unified API.

Thank you for your consideration!