API Reference barrel_embed v#2.6.0

Embedding coordinator

barrel_embed application callback module

Azure OpenAI embedding provider

AWS Bedrock embedding provider

CLIP image/text embedding provider

Cohere embedding provider

ColBERT late interaction embedding provider

FastEmbed embedding provider

Jina AI embedding provider

Local Python embedding provider

Mistral AI embedding provider

Ollama embedding provider

OpenAI embedding provider

One model process per model, shared by every database that uses it. A provider acquires its process by key; the pool starts it under barrel_embed_pool_sup on first use, counts references, and stops it when the last holder releases it. A process that dies is forgotten; the next acquire starts a new one.

Supervisor of the model processes shared through barrel_embed_pool. Children are temporary: the pool starts a new one on the next acquire after a crash.

Async Python port server with request multiplexing

Embedding provider behaviour definition

SPLADE sparse embedding provider

barrel_embed top level supervisor

Managed virtualenv for barrel_embed

Google Vertex AI embedding provider

Voyage AI embedding provider