Your evaluation workspace.
Build evaluations, run AI quality checks, compare outputs, inspect results, and track model performance from one place.
Evaluation activity
No runs yet
Create an evaluation and run your first dataset or live example to start seeing results here.
Create evaluation →Choose an evaluation
Guided Evaluation
Define your own criteria and evaluate AI outputs with a structured human review workflow.
AI Response Evaluation
Evaluate responses across factuality, instruction following, relevance, completeness, safety, and fluency.
Model Comparison
Compare two model outputs and identify which response performs better across your selected criteria.
Safety Evaluation
Test AI outputs for unsafe content, policy violations, harmful behavior, and severity.