AGC-Bench - ocw_connections: leaderboard
The "ocw_connections" 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 | Gemini 2.5 Flash | 1.19 | #245 |
| 2 | GPT-5.5 | 1.18 | #31 |
| 3 | GLM-5-Turbo | 1.15 | #119 |
| 4 | Hermes 4 405B | 1.11 | #334 |
| 5 | DeepSeek V3.1 NEX N1 | 1.1 | #327 |
| 6 | GLM-4.6 | 1.1 | #252 |
| 7 | GLM-4.7 | 1.09 | #192 |
| 8 | GLM-5 | 1.08 | #141 |
| 9 | trinity-large-preview | 1.08 | #582 |
| 10 | GLM-5.1 | 1.01 | #97 |
| 11 | GPT-5.4 | 1.01 | #79 |
| 12 | Hermes-3-Llama-3.1-405B | 1.01 | #510 |
| 13 | GPT-4o | 1 | #343 |
| 14 | Gemma 3 27B (IT) | 0.99 | #523 |
| 15 | Kimi K2.5 | 0.98 | #144 |
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-ocw-connections · How It Works · Data refreshed daily, snapshot 2026-10-20.