AGC-Bench - puntuguese: leaderboard
The "puntuguese" 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. 83 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | granite-4.0-h-micro | 2.49 | #1088 |
| 2 | Nova Lite (v1) | 1.48 | #830 |
| 3 | Mistral Nemo | 1.32 | #1085 |
| 4 | Llama 3.2 11B Instruct | 1.02 | #1126 |
| 5 | Llama 3.1 8B Instruct | 0.97 | #1035 |
| 6 | Gemini 2.5 Flash | 0.94 | #245 |
| 7 | Gemma 4 31B (IT) | 0.94 | #176 |
| 8 | Claude Sonnet 4 | 0.94 | #199 |
| 9 | DeepSeek V3.1 Terminus | 0.94 | #220 |
| 10 | Hermes-3-Llama-3.1-405B | 0.94 | #510 |
| 11 | GPT-5.1 | 0.88 | #135 |
| 12 | GPT-5.4 Mini | 0.76 | #212 |
| 13 | Gemma 4 26B A4B (IT) | 0.76 | #297 |
| 14 | GPT-4.1 Mini | 0.7 | #355 |
| 15 | GPT-4.1 | 0.68 | #248 |
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-puntuguese · How It Works · Data refreshed daily, snapshot 2026-10-20.