Skip to content
Get my AI grade

Published method

Arena text style-controlled human-preference ranking published 2026-07-21

Ranks official Arena model identifiers by the published style-controlled text overall rating.

arena-text-style-control-2026-07-21-fb100fd1a486-v1

Version

Jul 27, 2026

Published

1

Board

arena-text-preference

Method ID

Ranking rule

How the order is produced

official Arena rating descending

Refresh

When the method reruns

Each board shows the date of the snapshot it is built from.

Boards

Boards using this version

Every published board built with this method. Each opens on its latest snapshot.
BoardPrimary measureRowsConfidenceSnapshotBoard versionRefreshBoard
Arena text preferencearena-text-style-controlled-rating378100%Jul 21, 2026arena-text-style-control-2026-07-21-fb100fd1a486-v1Not statedOpen the board: Arena text preference

Versioned contract

Complete measurement ledger

These fields are stored with the methodology version. A changed contract must publish a new version before comparable movement can resume.

Arena

text

Method

Arena's official dataset describes style-controlled ratings; Arena changed its rating method from Elo to Bradley-Terry on January 9, 2024.

Source

arena-official

License

CC-BY-4.0

Category

overall

Identity

Official dataset model identifiers are retained exactly. Rank.ai does not merge aliases or silently replace model versions.

Universe

All 378 unique model identifiers in Arena's frozen text style-controlled overall release.

Versioning

Dataset Git revision, filtered row count, ordered row payload SHA-256, publish date, schema, and formulas are frozen.

Limitations

  • This is a human-preference family, not an absolute capability score.
  • Ratings are relative to the model cohort and Arena voting population.
  • The style-controlled text view does not measure every modality or task.
  • Confidence bounds and vote counts should be considered with point estimates.
  • A changed upstream release is suppressed until reviewed and pinned.

Missing Data

The release is suppressed unless every model has rating, lower and upper confidence bounds, variance, and vote count. Values are never imputed.

Publish Date

2026-07-21

Rank Formula

official Arena rating descending

Tie Breakers

  • official Arena upstream rank ascending
  • canonical entity slug ascending

Primary Metric

arena-text-style-controlled-rating

Dataset Revision

fb100fd1a48607efb93e28ba6610b641a7ac1bce

Style Controlled

Yes

Ranking Direction

descending

Canonical Rows Sha256

f558c95087051215765335e11c58acc20003c23386bafc2730346efca9a8aabc

Required Metrics Per Entity

  • arena-text-style-controlled-rating
  • arena-text-rating-lower-bound
  • arena-text-rating-upper-bound
  • arena-text-rating-variance
  • arena-text-vote-count

See how often AI names you.

Enter your website. In about two minutes, Rank.ai asks ChatGPT, Claude and Gemini 12 questions your customers ask and grades how often they name you.

  • Your grade out of 100How often AI names you, cites your site, and how high it ranks you.
  • Who gets namedEvery competitor in the answers, most named first.
  • The pages AI readsThe sources behind each answer.
  • Three fixesWhat to fix first, with a brief for the first article.