AGC-Bench - newyorker_humor: leaderboard
The "newyorker_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: Dataset z-score. Source: huggingface.co. Saturation forecast: Around June 2027. 82 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 | 1.06 | #31 |
| 2 | Claude Opus 4.6 | 1.01 | #63 |
| 3 | Claude Sonnet 4.5 | 0.95 | #142 |
| 4 | Claude Opus 4.7 | 0.95 | #49 |
| 5 | Claude Opus 4.5 | 0.95 | #83 |
| 6 | GPT-4.1 | 0.87 | #248 |
| 7 | Claude Sonnet 4 | 0.85 | #199 |
| 8 | Claude 3.7 Sonnet | 0.85 | #249 |
| 9 | GPT-5.4 | 0.82 | #79 |
| 10 | GPT-5.1 | 0.81 | #135 |
| 11 | GPT-4o | 0.77 | #343 |
| 12 | Claude Sonnet 4.6 | 0.71 | #90 |
| 13 | GLM-4.6 | 0.66 | #252 |
| 14 | Kimi K2.5 | 0.63 | #144 |
| 15 | GPT-5.4 Mini | 0.53 | #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-newyorker-humor · How It Works · Data refreshed daily, snapshot 2026-10-20.