Why Explicit Unrated States Beat Fake Benchmark Completeness
Many AI comparison tables silently backfill unknown speeds or intelligence scores with placeholder numbers. When an official provider omits max output caps, hardware chips, or third-party Artificial Analysis benchmarks, showing "Unrated" or "Not disclosed" preserves engineering trust and avoids distorting ROI formulas.
- Every verified active price cites its provider URL directly on /models/[slug] and /verification.
- Discrepancies against historical snapshot expectations are surfaced openly in data/verification-report.json.
- Any community-reported rate change submitted via /contact can be cross-checked against official documentation and logged cleanly in /changelog.
Inspecting Raw Data Programmatically
Developers and AI assistants can consume the normalized pricing matrix directly from /llm-price-index, /api/price-index, or the machine-readable /llms.txt and /llms-full.txt endpoints.