AGC-Bench - humor_transfer: leaderboard
The "humor_transfer" subtask of AGC-Bench. A task from AGC-Bench, evaluating models on open-ended creative and figurative-language generation — humor, metaphor, analogy, storytelling and brainstorming — typically graded by LLM-as-judge.
Metric: Dataset z-score. Source: huggingface.co. Saturation forecast: Not forecast. 82 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 | 0.97 | #79 |
| 2 | GPT-4.1 Mini | 0.97 | #355 |
| 3 | GPT-4.1 | 0.97 | #248 |
| 4 | GPT-5.5 | 0.97 | #31 |
| 5 | GPT-5.1 | 0.97 | #135 |
| 6 | GPT-5.4 Mini | 0.97 | #212 |
| 7 | GPT-5.4 Nano | 0.97 | #328 |
| 8 | Gemini 2.5 Flash | 0.88 | #245 |
| 9 | Gemma 3 12B (IT) | 0.88 | #664 |
| 10 | Gemma 3 4B (IT) | 0.88 | #992 |
| 11 | GPT-4o Mini | 0.88 | #599 |
| 12 | Claude Haiku 4.5 | 0.88 | #280 |
| 13 | Kimi K2.5 | 0.88 | #144 |
| 14 | DeepSeek V3.2 | 0.88 | #205 |
| 15 | Claude Opus 4.7 | 0.88 | #49 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=agc-bench-humor-transfer · How It Works · Data refreshed daily, snapshot 2026-10-20.