AGC-Bench - creatset: leaderboard
The "creatset" 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. 83 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3.2 3B Instruct | 1.63 | #1352 |
| 2 | Mistral Small 3.1 | 1.61 | #610 |
| 3 | Hermes 3 - Llama-3.1 70B | 1.61 | #927 |
| 4 | Llama 3 8B Instruct | 1.48 | #1129 |
| 5 | Llama 3.2 11B Instruct | 1.47 | #1126 |
| 6 | Llama 3.1 70B Instruct | 1.43 | #561 |
| 7 | Nova Micro (v1) | 1.39 | #831 |
| 8 | Llama 3.1 8B Instruct | 1.37 | #1035 |
| 9 | granite-4.0-h-micro | 1.37 | #1088 |
| 10 | Hermes 4 405B | 1.37 | #334 |
| 11 | Mistral Nemo | 1.3 | #1085 |
| 12 | Llama 4 Scout | 1.24 | #656 |
| 13 | Hermes 4 70B | 1.22 | #506 |
| 14 | pixtral-large-2411 | 1.18 | #478 |
| 15 | Llama 3.3 70B Instruct | 1.18 | #534 |
Interactive version: theaggregate.ai/benchmark?slug=agc-bench-creatset · How It Works · Data refreshed daily, snapshot 2026-10-20.