Overview

Your RAG evaluation workspace

Runs stored

–

Chat models

–

Embedding models

–

Strategies

–

Run	Kind	Config

New evaluation

Pick a strategy + models, paste your corpus & questions, get scored.

Generator model (provider/name)

Embedding model

Judge model (LLM-as-judge)

Metric preset

Strategy

Documents (one per line)

Questions (one per line) — prefix with 'Q::' to add reference answer

Head-to-head leaderboard across multiple configurations on the same corpus.

Documents (one per line)

Questions (one per line, '::ref' optional)

Configurations to compare

Every run executed through this server

ID	Kind	Config	Faithfulness	Relevance	Cost

Model ID	Type	Context	$ In/1M	$ Out/1M	Notes