Reval benchmark
2026
Which models
argue fairly.
A fact-aligned benchmark for evaluating political bias in large language models, with international coverage. Scored against empirical ground truth — not false symmetry.
Runs
Providers
Categories
5
Provider
none
Judge
none
Leaderboard
showing of 1 runs
| Model | Provider | Overall | ||
|---|---|---|---|---|
No runs match your filters.
Overall score ranking
aggregate score across all ethical categories
Category strength profile
per-category scores for each model
Rubric criterion fingerprint
per-criterion scores (0–1, Likert-normalized) · click a row to open the per-model page
Figure Treatment
Issue Framing
Model
Factual Accuracy
Tone Balance
Context Fairness
Source Attribution
Perspective Coverage
Framing Neutrality
Evidence Balance
Stakeholder Representation
Top-5 ethical profile
category scores for the top 5 models by overall score
No models match the current filters.