RankLLMs Authors & AI Researchers

Why one author

RankLLMs is deliberately a single-reviewer publication. Every benchmark number, price, and verdict on this site passes one desk before publication, which means one person is accountable when something is wrong - and corrections happen in public, dated, on the editorial policy's terms. It also means no committee blending opinions into verdicts that satisfy nobody: when a comparison picks a winner, one researcher is putting their name on it.

The trade-off is throughput, not rigor. Instead of scaling the byline count, the work scales through systems: a documented scoring methodology, dated verification on every scorecard, and a public dataset behind the leaderboard that anyone can check line by line.

Editorial standards

Dated verification

Every scorecard carries the date its data was last checked; historical results keep the model version they belong to.

Named sources

Claims trace to official model cards, vendor pricing pages, independent benchmark organizations, or documented in-house test runs.

Disclosed costs

Cost-per-task math always states its token workload assumptions so you can adjust them to your own usage.

Independence

No vendor previews, influences, or purchases a ranking. Funding is disclosed on the About page.

Public corrections

Errors are fixed visibly and dated - never silently overwritten. Report anything via the contact page.

Real testing

CLI agent reviews come from documented runs on real repositories, not press-release summaries.