Benchmark cards
Structured benchmark cases with scope, constraints, expected evidence, and reproducibility notes.
COG Eval
Evidence-based evaluation for AI engineering systems.
COG Eval may later be deployed as a separate product domain. This static page only previews the product direction.
Eval cards
Compact evaluation records for models, agents, workflows, and task-specific behavior.
Runner evidence
Execution traces, logs, artifacts, and review notes connected to each evaluation case.
Community retesting
A future pathway for comparing repeated runs and independent verification.
Workflow comparison
Side-by-side analysis for model, agent, and human-AI workflow performance.