Superlinked

Search inference: run encode, score, extract and generate models, plus embeddings and transcription

Inference: chat, completions, embeddings, rerank, audio transcription, plus model listing and runtime pool coordination