AGC-Bench - pollux_creativity: leaderboard
The "pollux_creativity" 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 March 2027. 83 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 | 1.22 | #79 |
| 2 | Gemma 4 31B (IT) | 0.98 | #176 |
| 3 | Gemma 4 26B A4B (IT) | 0.98 | #297 |
| 4 | GPT-5.4 Mini | 0.92 | #212 |
| 5 | Kimi K2.6 | 0.91 | #104 |
| 6 | Claude Opus 4.6 | 0.9 | #63 |
| 7 | Claude Sonnet 4.5 | 0.9 | #142 |
| 8 | Qwen 3 Next 80B A3B Instruct | 0.89 | #368 |
| 9 | DeepSeek V3.1 Terminus | 0.87 | #220 |
| 10 | Claude 3.7 Sonnet | 0.86 | #249 |
| 11 | GLM-5.1 | 0.86 | #97 |
| 12 | Mistral Medium 3.1 | 0.8 | #408 |
| 13 | Kimi K2.5 | 0.78 | #144 |
| 14 | GPT-5.5 | 0.77 | #31 |
| 15 | Claude Opus 4.7 | 0.77 | #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-pollux-creativity · How It Works · Data refreshed daily, snapshot 2026-10-20.