omniroute-embeddings

What problem does it solve?

It solves the problem of generating high-quality vector embeddings reliably when a specific embedding provider model is slow, rate-limited, or unavailable.

Core Features & Use Cases

• Embeddings via OmniRoute: Creates embeddings using the /v1/embeddings interface compatible with OpenAI-style requests.
• Auto-fallback across providers: Routes the embedding request across multiple embedding providers to keep responses flowing.
• RAG and similarity search ready: Produces vectors suitable for retrieval-augmented generation, semantic search, clustering, and other similarity workflows.

Use it when you need embeddings for a RAG pipeline and want resilience by automatically switching between models such as text-embedding-3-large, Voyage, Cohere, Gemini, or Jina.

Quick Start

Send a POST request to $OMNIROUTE_URL/v1/embeddings with Authorization: Bearer $OMNIROUTE_KEY and a JSON body specifying model, input, and encoding_format.