selecting-embedding-model

What problem does it solve?

This Skill helps you choose an embedding model using evidence from your own domain instead of public leaderboards, so you do not pick a model that looks strong in general but fails on your retrieval task.

Core Features & Use Cases

• Compares candidates on intrinsic similarity ranking, extrinsic retrieval quality, and operational cost, latency, and index size.
• Handles new RAG builds, semantic search projects, and embedding swaps when models look similar or trade-offs are unclear.
• Rejects leaderboard-only recommendations and forces domain-specific evaluation before making a choice.

Quick Start

Ask the skill to compare your candidate embedding models on a domain-specific golden set and return a recommendation table with quality, cost, latency, and index-size trade-offs.