Moneda Bench leaderboard
Moneda CiteBench
Whether the sources a system cites say what it claims. The system writes a short research note on a company, with sources, and every linked claim is checked against the page it links to.
First results
Coming
First results coming
No results yet. This is how the board will work once they are in.
Each system gets the same plain retail prompt per company. Claims are split into numeric and descriptive. The judge is checked against people on a held-out set before the rubric is tuned, and we report agreement next to the scores. Below 0.7 agreement the board is labelled experimental and not ranked.
What it will report
- Factual share (higher is better): Share of linked claims whose source supports them. The headline number.
- Unlinked share (lower is better): Share of claims that carry no source at all.