AGC-Bench - Humor: leaderboard
The "Humor" 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: JRT z-score. Source: huggingface.co. Saturation forecast: Around April 2027. 83 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.7 | 1.09 | #49 |
| 2 | GPT-5.5 | 0.86 | #31 |
| 3 | GPT-5.4 | 0.76 | #79 |
| 4 | Gemma 4 31B (IT) | 0.69 | #176 |
| 5 | GLM-5-Turbo | 0.65 | #119 |
| 6 | Claude Sonnet 4.5 | 0.63 | #142 |
| 7 | Gemma 4 26B A4B (IT) | 0.62 | #297 |
| 8 | GLM-4.7 | 0.61 | #192 |
| 9 | GPT-5.1 | 0.61 | #135 |
| 10 | DeepSeek V3.1 Terminus | 0.59 | #220 |
| 11 | Claude 3.7 Sonnet | 0.58 | #249 |
| 12 | Claude Haiku 4.5 | 0.58 | #280 |
| 13 | Claude Opus 4.5 | 0.58 | #83 |
| 14 | Kimi K2 0905 | 0.56 | #288 |
| 15 | GPT-5.4 Mini | 0.56 | #212 |
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 · How It Works · Data refreshed daily, snapshot 2026-10-20.