A reproducible local LLM benchmarking platform with versioned datasets, deterministic scoring, and auditable reports.
Terminal score: 0–1000 raw, weighted across 4 dimensions. Public score: 0–10 normalized (shown in the 30-day stars chart above).
