AGC-Bench - science_analogies: leaderboard
The "science_analogies" 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. 81 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4o | 2.5 | #343 |
| 2 | Hermes 4 405B | 2.45 | #334 |
| 3 | minimax-m2-her | 2.23 | |
| 4 | GPT-4o Mini | 2.15 | #599 |
| 5 | Hermes 4 70B | 2 | #506 |
| 6 | trinity-large-preview | 1.94 | #582 |
| 7 | Hermes-3-Llama-3.1-405B | 1.58 | #510 |
| 8 | Hermes 3 - Llama-3.1 70B | 1.5 | #927 |
| 9 | GLM-5.1 | 1.15 | #97 |
| 10 | Qwen 3 14B | 1.03 | #537 |
| 11 | jamba-large-1.7 | 0.89 | #525 |
| 12 | Gemma 4 31B (IT) | 0.79 | #176 |
| 13 | ERNIE 4.5 300B A47B | 0.63 | #458 |
| 14 | GPT-5.5 | 0.57 | #31 |
| 15 | Qwen 3.5 Flash (02-23) | 0.55 | #185 |
Interactive version: theaggregate.ai/benchmark?slug=agc-bench-science-analogies · How It Works · Data refreshed daily, snapshot 2026-10-20.