PunEval
Models recognize, explain and generate English puns. The evaluation checks recognition against labels and compares explanations with human annotations.
The authors publish recognition results for eight models. A later EMNLP study tested seven language models on its revised 1,341-item PunEval detection split.
- Made by
- Fudan University
- Released
- April 2024
- Kind
- Language reasoning
- Level
- Everyday
- Scored
- Graded by a model
- Answers
- Open-ended
- Tasks
- All public
- Measures
- Capability
- Subject
- Languages and literature
- Language
- English
- Interaction
- Single turn
- Items from
- Written by crowdworkers
- Refresh
- One fixed set
- Run by
- Anyone
- Based on
- ExPUNations
- Strongest model tested
- Claude 3 Opus #173 on the capability index
- Last model added
- February 2024
- Scores refreshed
- November 20th, 2026
2 models share the top score of 1.00 among models released by February 2024 2 models share the top score of 1.00 among models released by February 2024
Best score per entry from the benchmark's own leaderboard, read on November 20th, 2026. The tasks use different metrics, so the headline is the homographic pun recognition true positive rate with the pun-biased prompt. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.
The newest model with a score here came out in February 2024, 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 GPT-3.5 Turbo Nov 2023, Zero-shot, without definition, chain of thought 1.00
- 2
Gemini 1.0 Pro Zero-shot, without definition, chain of thought 1.00
- 3 vicuna-7b-v1.5 Zero-shot, with definition, direct answer 1.00
- 4 llama-2-7b-chat Six examples, without definition, direct answer 1.00
- 5 openchat-3.5-0106 Zero-shot, without definition, chain of thought 1.00
- 6 GPT-4 Turbo Nov 2023, Zero-shot, without definition, chain of thought 0.99
- 7
Mistral-7B-Instruct-v0.2 Zero-shot, without definition, chain of thought 0.99
- 8 Claude 3 Opus Zero-shot, without definition, direct answer 0.99
Source: Official PunEval leaderboard
- 1 GPT-3.5 Turbo Nov 2023, Zero-shot, without definition, chain of thought 1.00
- 2
Gemini 1.0 Pro Zero-shot, without definition, chain of thought 1.00
- 3 GPT-4 Turbo Nov 2023, Zero-shot, without definition, chain of thought 0.99
- 4
Mistral-7B-Instruct-v0.2 Zero-shot, without definition, chain of thought 0.99
- 5 Claude 3 Opus Zero-shot, without definition, direct answer 0.99
Source: Official PunEval leaderboard
| # | Entry | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-3.5 Turbo Nov 2023 | 1.00 | Zero-shot, without definition, chain of thought | Nov 2023 | Official leaderboard |
| 2 | | 1.00 | Zero-shot, without definition, chain of thought | Dec 2023 | Official leaderboard |
| 3 | vicuna-7b-v1.5 | 1.00 | Zero-shot, with definition, direct answer | Official leaderboard | |
| 4 | llama-2-7b-chat | 1.00 | Six examples, without definition, direct answer | Official leaderboard | |
| 5 | openchat-3.5-0106 | 1.00 | Zero-shot, without definition, chain of thought | Official leaderboard | |
| 6 | GPT-4 Turbo Nov 2023 | 0.99 | Zero-shot, without definition, chain of thought | Nov 2023 | Official leaderboard |
| 7 | | 0.99 | Zero-shot, without definition, chain of thought | Dec 2023 | Official leaderboard |
| 8 | Claude 3 Opus | 0.99 | Zero-shot, without definition, direct answer | Feb 2024 | Official leaderboard |
| # | Model | Score | Setting | Date | Source |
|---|---|---|---|---|---|
| 1 | GPT-3.5 Turbo Nov 2023 | 1.00 | Zero-shot, without definition, chain of thought | Nov 2023 | Official leaderboard |
| 2 | | 1.00 | Zero-shot, without definition, chain of thought | Dec 2023 | Official leaderboard |
| 3 | GPT-4 Turbo Nov 2023 | 0.99 | Zero-shot, without definition, chain of thought | Nov 2023 | Official leaderboard |
| 4 | | 0.99 | Zero-shot, without definition, chain of thought | Dec 2023 | Official leaderboard |
| 5 | Claude 3 Opus | 0.99 | Zero-shot, without definition, direct answer | Feb 2024 | Official leaderboard |
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: Official PunEval leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-3.5 Turbo Nov 2023 | Zero-shot, without definition, chain of thought | 1.00 | |||
| Zero-shot, without definition, chain of thought | 1.00 | ||||
| GPT-3.5 Turbo Nov 2023 | Zero-shot, with definition, chain of thought | 1.00 | |||
| Zero-shot, without definition, direct answer | 1.00 | ||||
| Zero-shot, with definition, chain of thought | 1.00 | ||||
| Six examples, without definition, direct answer | 1.00 | ||||
| Six examples, with definition, direct answer | 1.00 | ||||
| Zero-shot, with definition, direct answer | 1.00 | ||||
| GPT-4 Turbo Nov 2023 | Zero-shot, without definition, chain of thought | 0.99 | |||
| GPT-4 Turbo Nov 2023 | Zero-shot, with definition, chain of thought | 0.99 | |||
| Zero-shot, without definition, chain of thought | 0.99 | ||||
| Six examples, without definition, chain of thought | 0.99 | ||||
| Six examples, with definition, chain of thought | 0.99 | ||||
| GPT-3.5 Turbo Nov 2023 | Six examples, without definition, chain of thought | 0.99 | |||
| GPT-4 Turbo Nov 2023 | Six examples, with definition, chain of thought | 0.99 | |||
| GPT-4 Turbo Nov 2023 | Six examples, without definition, chain of thought | 0.99 | |||
| GPT-3.5 Turbo Nov 2023 | Zero-shot, without definition, direct answer | 0.99 | |||
| GPT-3.5 Turbo Nov 2023 | Six examples, with definition, chain of thought | 0.99 | |||
| Claude 3 Opus | Zero-shot, without definition, direct answer | 0.99 | |||
| GPT-4 Turbo Nov 2023 | Zero-shot, without definition, direct answer | 0.99 | |||
| GPT-4 Turbo Nov 2023 | Six examples, with definition, direct answer | 0.99 | |||
| GPT-4 Turbo Nov 2023 | Six examples, without definition, direct answer | 0.99 | |||
| GPT-3.5 Turbo Nov 2023 | Six examples, without definition, direct answer | 0.98 | |||
| Claude 3 Opus | Zero-shot, with definition, direct answer | 0.98 | |||
| Claude 3 Opus | Six examples, with definition, direct answer | 0.98 | |||
| Claude 3 Opus | Six examples, with definition, chain of thought | 0.98 | |||
| GPT-3.5 Turbo Nov 2023 | Zero-shot, with definition, direct answer | 0.97 | |||
| GPT-3.5 Turbo Nov 2023 | Six examples, with definition, direct answer | 0.97 | |||
| GPT-4 Turbo Nov 2023 | Zero-shot, with definition, direct answer | 0.97 | |||
| Zero-shot, with definition, chain of thought | 0.91 | ||||
| Zero-shot, without definition, direct answer | 0.87 | ||||
| Six examples, without definition, chain of thought | 0.73 | ||||
| Six examples, with definition, chain of thought | 0.73 | ||||
| Six examples, without definition, direct answer | 0.68 | ||||
| Six examples, with definition, direct answer | 0.57 | ||||
| Zero-shot, with definition, direct answer | 0.54 |
Scores at each effort level
Models the leaderboard ran at two or more effort levels. We don’t have enough cost data here to say what the extra effort costs.
Source: Official PunEval leaderboard
| Model | Effort | Harness | Score | Cost per task | Output tokens |
|---|---|---|---|---|---|
| GPT-3.5 Turbo Nov 2023 | Zero-shot, without definition, chain of thought | 1.00 | |||
| Zero-shot, without definition, chain of thought | 1.00 | ||||
| GPT-3.5 Turbo Nov 2023 | Zero-shot, with definition, chain of thought | 1.00 | |||
| Zero-shot, without definition, direct answer | 1.00 | ||||
| Zero-shot, with definition, chain of thought | 1.00 | ||||
| Six examples, without definition, direct answer | 1.00 | ||||
| Six examples, with definition, direct answer | 1.00 | ||||
| Zero-shot, with definition, direct answer | 1.00 | ||||
| GPT-4 Turbo Nov 2023 | Zero-shot, without definition, chain of thought | 0.99 | |||
| GPT-4 Turbo Nov 2023 | Zero-shot, with definition, chain of thought | 0.99 | |||
| Zero-shot, without definition, chain of thought | 0.99 | ||||
| Six examples, without definition, chain of thought | 0.99 | ||||
| Six examples, with definition, chain of thought | 0.99 | ||||
| GPT-3.5 Turbo Nov 2023 | Six examples, without definition, chain of thought | 0.99 | |||
| GPT-4 Turbo Nov 2023 | Six examples, with definition, chain of thought | 0.99 | |||
| GPT-4 Turbo Nov 2023 | Six examples, without definition, chain of thought | 0.99 | |||
| GPT-3.5 Turbo Nov 2023 | Zero-shot, without definition, direct answer | 0.99 | |||
| GPT-3.5 Turbo Nov 2023 | Six examples, with definition, chain of thought | 0.99 | |||
| Claude 3 Opus | Zero-shot, without definition, direct answer | 0.99 | |||
| GPT-4 Turbo Nov 2023 | Zero-shot, without definition, direct answer | 0.99 | |||
| GPT-4 Turbo Nov 2023 | Six examples, with definition, direct answer | 0.99 | |||
| GPT-4 Turbo Nov 2023 | Six examples, without definition, direct answer | 0.99 | |||
| GPT-3.5 Turbo Nov 2023 | Six examples, without definition, direct answer | 0.98 | |||
| Claude 3 Opus | Zero-shot, with definition, direct answer | 0.98 | |||
| Claude 3 Opus | Six examples, with definition, direct answer | 0.98 | |||
| Claude 3 Opus | Six examples, with definition, chain of thought | 0.98 | |||
| GPT-3.5 Turbo Nov 2023 | Zero-shot, with definition, direct answer | 0.97 | |||
| GPT-3.5 Turbo Nov 2023 | Six examples, with definition, direct answer | 0.97 | |||
| GPT-4 Turbo Nov 2023 | Zero-shot, with definition, direct answer | 0.97 | |||
| Zero-shot, with definition, chain of thought | 0.91 | ||||
| Zero-shot, without definition, direct answer | 0.87 | ||||
| Six examples, without definition, chain of thought | 0.73 | ||||
| Six examples, with definition, chain of thought | 0.73 | ||||
| Six examples, without definition, direct answer | 0.68 | ||||
| Six examples, with definition, direct answer | 0.57 | ||||
| Zero-shot, with definition, direct answer | 0.54 |
Number one over time
Lab claims count from the day the card came out, independent runs from the model’s release.
Source: Official PunEval leaderboard
Scores over time
All the scores we have, one dot each, so a model tested at several settings appears more than once.
Source: Official PunEval 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 Knowledge
- MMLU-Pro In use Harder 10-option version of MMLU with reasoning-focused questions across 14 subjects.
Intern-S2-397B 89.8%
- MMLU In use 57-subject multiple-choice exam questions from elementary to professional level. Scored by accuracy.
Seed1.8 92.3%
- SuperGPQA In use 26,529 graduate-level multiple-choice questions across 285 disciplines. Accuracy.
Qwen 3.6 Max (Preview) 73.9%
- WideSearch In use 200 bilingual broad info-seeking tasks: agent collects many atomic facts into a table. Scored by success rate and row/item F1.
Tencent Hy4 preview 83.9%
- BBEH micro average In use Models solve harder replacements for each of the 23 BIG-Bench Hard tasks, 4,520 examples in total. The score is the share of all examples answered correctly.
Gemma 4 31B IT 74.4%
- HLE-Diamond In use Models answer 1,000 cleaned expert questions from Humanity's Last Exam, 500 on reasoning and 500 on knowledge, closed book. Graded on exact answers. Claude Fable 5.2 77.9%
Data audit
8 of 9 checks pass, 1 warning. Look first at the Newest score check. The newest came 2 years ago, in February 2024, for Claude 3 Opus. Every benchmark
Catalog record 2 pass
- Pass Task access All tasks are public.
- Pass Confidence Checked against a primary source.
Scores 1 warning 2 pass 2 notes
- Pass Independent runs Scores from the official leaderboard (8 entries).
- Pass Models scored 5 models we track, plus 3 other entries on its leaderboard.
- Warning Newest score The newest came 2 years ago, in February 2024, for Claude 3 Opus.
- Note Frontier models None of the top 10 has a score. The strongest tested is Claude 3 Opus (#173).
- Note Headroom Scored as a plain number, with no fixed ceiling.
Leaderboard import 3 pass
- Pass Importer We read the board on github.com: 36 rows for models we track and 3 other entries at the last read.
- Pass Last read November 20th, 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.