Per-DOC
Models use a reader’s previous comparison to predict which of two new plots the reader prefers on each of five criteria.
PREFINE reran the authors’ test on unseen reviewers. PerSE-Llama3-8B averaged 63.3% accuracy across the five criteria.
- Made by
- Carnegie Mellon University, Meta AI and UC Berkeley
- Released
- October 2023
- Kind
- Taste and style
- Level
- Everyday
- Scored
- Checked automatically
- Answers
- Multiple choice
- Tasks
- All public
- Measures
- Capability
- Language
- English
- Interaction
- Single turn
- Items from
- Written by crowdworkers
- Refresh
- One fixed set
- Run by
- Anyone
- Strongest model tested
-
Llama 2-70B #210 on the capability index
- Last model added
- July 2023
- Scores refreshed
- November 21st, 2026
PERSE-13b tops the leaderboard with 0.34% Llama 2-70B leads with 0.21% among models released by July 2023
Best score per entry from the benchmark's paper, read on November 21st, 2026. A paper’s table doesn’t add newer models. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is Llama 2-70B, at 0.21%. This view shows only models we track, each at its best.
The newest model with a score here came out in July 2023, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:
- GPT-6.2 Astra
- Claude Fable 5.2
-
Gemini 4 Carbon
- GPT-6 Astra
- and 16 more
- 1 PERSE-13b 0.34%
- 2 GPT-4 (snapshot unspecified) 0.32%
- 3 PERSE-7b 0.31%
- 4
Llama 2-70B 0.21%
- 5
Llama 2-13B 0.17%
- 6
Llama 2-7B 0.15%
Source: The Per-DOC paper
- 1
Llama 2-70B 0.21%
- 2
Llama 2-13B 0.17%
- 3
Llama 2-7B 0.15%
Source: The Per-DOC paper
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | PERSE-13b | 0.34% | Official leaderboard | ||
| 2 | GPT-4 (snapshot unspecified) | 0.32% | Official leaderboard | ||
| 3 | PERSE-7b | 0.31% | Official leaderboard | ||
| 4 | | 0.21% | Jul 2023 | Official leaderboard | |
| 5 | | 0.17% | Jul 2023 | Official leaderboard | |
| 6 | | 0.15% | Jul 2023 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | | 0.21% | Jul 2023 | Official leaderboard | |
| 2 | | 0.17% | Jul 2023 | Official leaderboard | |
| 3 | | 0.15% | Jul 2023 | Official leaderboard |
Rankings by part
We read 2 parts from the benchmark’s paper: Spearman correlation and Kendall correlation. Each ranks models on one part of the test. Llama 2-70B leads both. Arrows show how far a model moved from its overall rank among the models we track.
Spearman correlation
3 models we track have a score here.
- 1
Llama 2-70B 0.23%
- 2
Llama 2-13B 0.18%
- 3
Llama 2-7B 0.12%
Kendall correlation
3 models we track have a score here.
- 1
Llama 2-70B 0.18%
- 2
Llama 2-13B 0.15%
- 3
Llama 2-7B 0.09%
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: The Per-DOC paper
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official Per-DOC leaderboard
No lab model card reports it yet
We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark.
More in Writing and taste
- Design Bench (Moonshot) In use Models make graphic designs from direct prompts or few-shot reference images. An automated grader rates the results as Elo scores. Kimi K3.1 Pro 957
- WritingBench Fading Models complete 1,000 real-world writing requests across 6 domains and 100 subdomains. A judge model scores each reply on five query-specific criteria. GPT-5 83.9
- Creative Writing (MiniMax in-house) Only in system cards Models write creative pieces for real Hailuo AI user requests. Automatic assessments score the writing.
MiniMax-Text-01 81.3%
- DeepSeek Chinese functional writing evaluation Only in system cards Models answer 3,170 everyday Chinese writing prompts such as reports, emails and speeches. Pairwise judging against a baseline model gives a win rate.
DeepSeek-V4-Pro 62.6%
- Slide generation human evaluation (Z.ai) Only in system cards People compare HTML slide decks from two models on content, layout and visual quality. Metric is overall win rate. GLM-5 67.5%
- Task-dependent output homogenization evaluation Only in system cards Models answer prompts several times. A judge scores functional diversity where it is wanted and consistency where one answer is right.
Muse Spark 0.21
Data audit
9 of 10 checks pass, 1 warning. Look first at the Newest score check. The newest came 3 years ago, in July 2023, for Llama 2-13B. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 3 pass 1 note
- Pass Independent runs Scores from the benchmark’s paper (6 entries).
- Pass Models scored 3 models we track, plus 3 other entries on its leaderboard.
- Warning Newest score The newest came 3 years ago, in July 2023, for Llama 2-13B.
- Note Frontier models None of the top 10 has a score. The strongest tested is Llama 2-70B (#210).
- Pass Headroom The best is 0.21%, under 85%, so it still separates models.
Leaderboard import 3 pass
- Pass Importer We read the board on aclanthology.org: 3 rows for models we track and 3 other entries at the last read.
- Pass Last read November 21st, 2026.
- Pass Read worked The last read worked.
Consistency 1 pass 1 note
- Note Lab claims No model has both a lab number and an independent run.
- Pass Known issues No published audit found problems.