AGC-Bench - conceptual_design: leaderboard
The "conceptual_design" 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 June 2027. 83 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 Next 80B A3B Instruct | 2.22 | #368 |
| 2 | Kimi K2 0905 | 2.02 | #288 |
| 3 | Kimi K2.6 | 1.57 | #104 |
| 4 | GLM-4.6 | 1.41 | #252 |
| 5 | GLM-5-Turbo | 1.36 | #119 |
| 6 | DeepSeek V3.1 Terminus | 1.3 | #220 |
| 7 | Kimi K2.5 | 1.24 | #144 |
| 8 | Qwen 3.5 Flash (02-23) | 1.23 | #185 |
| 9 | GLM-5.1 | 1.14 | #97 |
| 10 | DeepSeek V3 Chat | 1.03 | #447 |
| 11 | Nemotron 3 Super | 0.98 | #335 |
| 12 | DeepSeek V3.1 | 0.97 | #264 |
| 13 | DeepSeek V3.1 NEX N1 | 0.96 | #327 |
| 14 | GLM-4.7 | 0.89 | #192 |
| 15 | GLM-5 | 0.85 | #141 |
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-conceptual-design · How It Works · Data refreshed daily, snapshot 2026-10-20.