APUN-Bench (Synthetic) - Homophonic Pun Inference: leaderboard
Metric: Homophonic pun-inference accuracy (%), the predicted alternative word exactly matching the gold word, on the synthetic set (3,541 SemEval-2017 pun and non-pun sentences voiced with Parler-TTS); temperature 0; higher is better. Source: arxiv.org. Saturation forecast: Estimated already saturated. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.0 Flash | 92.11 | #331 |
| 2 | MERaLiON-2-10B | 79.93 | #567 |
| 3 | Qwen2.5-Omni-7B | 60.19 | #601 |
| 4 | Qwen2-Audio-7B-Instruct | 26.97 | #859 |
| 5 | SALMONN-13B | 25.32 | #1237 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #8 Claude Fable 5.1.
Interactive version: theaggregate.ai/benchmark?slug=apun-bench-synthetic-homophonic-pun-inference · How It Works · Data refreshed daily, snapshot 2026-10-11.